Pharmaceutical Quality Control of Complex Botanicals Dr Peter Gibson Technical Director Porton Down Science Park, Salisbury, SP4 0JQ, United Kingdom
GW Pharmaceuticals To be the global leader in prescription cannabinoid medicines, through the development of pharmaceutical products which address clear unmet patient needs
Single Chemical versus Botanical Single Molecule Single identified active Full characterisation Limited number of related substances Botanical One or more actives May not be identified Wide range of components FID1 A, (CHEMSTOR\3940163\A417843\APB00021.D) FID1 A, (CHEMSTOR\3940163\A417843\APB00021.D) count count 4800 Cannabinoids THC 4800 4600 4600 4400 Sesqui terpenes Di terpenes 4400 4200 Mono terpenes Tri terpenes & Waxes 4200 4000 4000 3800 3800 3600 3600 0 10 20 30 40 0 10 20 30 40 mi m
FDA Botanical Guidelines (2004) For Phase III prior to NDA submission: Well characterized Batch to batch consistency Multiple fingerprints Qualitatively and quantitatively comparable
Lead Product Sativex Oromucosal Spray Spasticity in multiple sclerosis Pain in advanced cancer Neuropathic pain Approved in 24 countries IND in the US Botanical extracts from specific Cannabis sativa plants THC 27mg/ml CBD 25mg/ml
Sativex Production Overview THC BRM Botanical Raw Material (BRM) CBD BRM THC BDS Botanical Drug Substance (BDS) CBD BDS Botanical Drug Product (BDP) Sativex
Cannabis Production Weeks 1 to 3
Cannabis Production Weeks 4 to 11
Cannabis Production Weeks 4 to 11
BRM to BDS THC BRM CBD BRM Milling Decarboxylation CO 2 Extraction Partial Purification Isolation THC BDS CBD BDS
BDS to Sativex THC BDS CBD BDS Bulk Solution Filling Packaging Sativex Finished Product
BDS Characterization Cannabinoids Non cannabinoids 70 to 75% w/w 25 to 30% w/w
BDS Total Components Identified 400 500 compounds present in the BDS n = 73 n = 79
Specification Quantitative 10 cannabinoids 14 non cannabinoid components Class totals Qualitative Chromatogram comparison Routine use of 6 methods
Qualitative Compare Chromatograms
Overview 4 methods Cannabinoids Terpenes Triglycerides Sterols & Triterpenes
Basis for Interpretation PC2 PC3 PC1
System Approach PCA Model Cannabinoids PCA Model Terpenes PCA Model Sterols PCA Model Triglycerides Compare new batch Compare new batch Compare new batch Compare new batch Pass/Warn/Fail Pass/Warn/Fail Pass/Warn/Fail Pass/Warn/Fail Pass/Warn/Fail Qualitative Fit of Batch
Qualitative Assessment Process Fingerprinting Automate batch assessment using pass warn fail system
Alignment issue Instrument Column Mobile Phase
Raw and Aligned Data Raw Data Aligned
Model from Aligned Profiles Each evaluation has two qualifiers in model consistency Mahalanobis distance out of model variation Sample Residual PCA Scores Warning limit (95% confidence interval) Failure limit (99% confidence interval)
CBD BDS Terpene Analysis Outliers in bold 23
CBD BDS Cannabinoid Analysis Outliers in bold
THC BDS Models Cannabinoids Terpene s Sterols Triglyceride s
CBD BDS Models Cannabinoids Terpene s Sterols Triglyceride s
Sativex Models Cannabinoids Terpenes
Model Development PCA Model Cannabinoids PCA Model Terpenes PCA Model Sterols PCA Model Triglycerides Compare new batch Compare new batch Compare new batch Compare new batch Pass/Warn/Fail Pass/Warn/Fail Pass/Warn/Fail Pass/Warn/Fail Pass/Warn/Fail Qualitative Fit of Batch
QC Analysis Schema the Profiler THC or CBD or Sativex Cannabinoids (LC) Sterols (GC) Terpenes (GC) Triglycerides (LC) CDF CDF CDF CDF Align Signals CDF CDF CDF CDF Compare Fingerprint to model MD & Q Report Scoresheet
Batch Scoring For MD, Q if metric < 95% cutoff, score = 0 if metric > 95% cutoff, score = 1 if metric > 99% cutoff, score = 3 Cannabinoids Fail if (primary) score >= 3 Warn if score > 0 Non cannabinoids Fail if composite (secondary) score >= 9 Warn if composite score >= 3
Investigate Warn/Fail CBN THCV CBC
Summary Routine QC Procedure Based on strong data set Demonstrates can identify batches Absence of expected constituents Presence of the unexpected peak Variation in the expected relative abundance Chromatographic changes beyond the scope of the training set
Thanks to: Infometrics Inc Brian Rohrback Scott Ramos Analytical R&D team at GW Pharma Alan Sutton Emma Lennon James Dean Hughes