1. Addressing the analytical hurdles associated with Artemisia annua extracts Josh L Pilkington, Chris Preston and Rachel L Gomes Department of Chemical and Environmental Engineering, University of Nottingham, University Park, Nottingham NG7 2RD

2. Malaria  247 million cases in 2008  Almost one million deaths annually  Concentrated in sub-Saharan Africa

3. Artemisinin  Precursor to Artemisinin-based Combination Therapies (ACTs), considered the most effective treatment against Plasmodium falciparum 1  Primarily obtained through solvent extraction from the leaves of Artemisia annua, grown in temperate climates  Total synthesis is possible but prohibitively expensive  Production (Africa and Asia) requires improvements in efficiency (currently 50-60%) and cost-effectiveness 1 WHO (2011) Global Plan for Artemisinin Resistance Containment [online] http://www.who.int/malaria/publications/atoz/artemisinin_resistance_ containment_2011.pdf [accessed 11/07/2011]

4. Artemisinin Solvent Extraction Solvent extractionConcentration Solvent decanted Crystals harvested (Crude artemisinin) Purification through multiple ethanol re-crystallisations

5. Artemisinin Detection  In order to assess the efficiency of each processing stage, robust analytical techniques are required to determine the relationship between processing parameters and overall processing efficiency  Most methods in the literature only analyse high purity artemisinin and are unsuited to earlier processing stages  Analytical techniques involving LC-MS or NMR require capital investments or expertise that are not available to producers

6. Liquid Chromatography (LC)  HPLC-UV is likely to demonstrate the most favourable balance between equipment cost, simplicity of operation and accuracy  Lapkin et al. [2] recently undertook a substantial investigation into various different HPLC methods: -Compared various detection methods -Analysed extracts produced with different solvents -Concluded that HPLC-UV was not suitable for A. annua extracts due to the high presence of impurities. 2 Lapkin, A. A., Walker, A., Sullivan, N., Khambay, B., Mlambo, B., Chemat, S. (2009) Development of HPLC analytical protocols for quantification of artemisinin in biomass and extracts. Journal of Pharmaceutical and Biomedical Analysis, 49, 908-915.

7. Aims and Objectives  Develop a functional HPLC-UV analytical method to analyse the artemisinin content of crude A. annua extracts -Identify problems with Lapkin method -Overcome difficulties with impurities -Evaluate against a range of extraction solvents  Utilise the HPLC-UV method to compare the extraction efficiency of artemisinin using different solvents - Inform on optimum processing conditions

8. Producing Extracts  Followed the procedure detailed in the recent publication by Lapkin et al. [2]  Static, 4 hour extractions of 10g of leaf with 100ml solvent (ethyl acetate, hexane, ethanol and a 95:5 hexane:ethyl acetate mixture) in triplicate  An additional, agitated extraction with ethyl acetate was performed for comparison  Extracts strained through muslin fabric to separate leaves from extract (miscella) before vacuum filtration 2 Lapkin, A. A., Walker, A., Sullivan, N., Khambay, B., Mlambo, B., Chemat, S. (2009) Development of HPLC analytical protocols for quantification of artemisinin in biomass and extracts. Journal of Pharmaceutical and Biomedical Analysis, 49, 908-915.

9. Repeating Lapkin Method  Co-elution of artemisinin and impurities was observed for all but the ethanol extracts  Greater separation of impurities from artemisinin required for quantification

10. New HPLC-UV Method  New method has a longer duration but provides good separation of artemisinin from impurities  Limit of Quantification (LOQ) at λ=210nm was found to be 12μg/ml  Extract concentrations were between 430 and 920μg/ml depending on the extraction solvent, so artemisinin could be readily quantified

11. Observation  To analyse the artemisinin content of an extract, the extraction solvent (e.g. hexane) is usually evaporated off to leave a dry residue  The residue is then re-dissolved into another solvent (often acetonitrile) to be analysed by HPLC. This is sample reconstitution  During sample reconstitution, it was observed that most of the residue remains in the vial. Any solid residue remaining after reconstitution cannot be detected Could artemisinin remain trapped inside the residue after reconstitution? This would lead to an underestimate of the actual artemisinin content of the extract

12. Extract Reconstitution Aliquots of each extract were evaporated to dryness and the residue weight recorded. Four methods of reconstitution in acetonitrile were then undertaken: 1. Thirty seconds on vortex (2800rpm) 2. Ten minutes mixing (350rpm) 3. Thirty minutes mixing 4. Twenty four hours mixing 123 4

13. Results: Effect of reconstitution method

14. Summary: Effect of reconstitution method  Increasing the duration of reconstitution increased the total amount of detectable artemisinin in all extract residues  The impurity profile of extracts is also affected by reconstitution duration  The amount of artemisinin reconstituted after 24 hours is the same as after 48 hours, indicating the maximum amount of artemisinin attainable has been achieved

15. Results: Effect of reconstitution method

16. Hypothesis - Mechanism For 30 seconds of reconstitution, acetonitrile is only able to contact artemisinin contained on the surface of the extract residue Extract Residue Reconstituted Acetonitrile KEY TRIPLICATE VIALS

17. Hypothesis - Mechanism For intermediate reconstitution durations (10 and 30 minutes), the penetration of acetonitrile into wax is variable. Extract Residue Reconstituted Acetonitrile KEY TRIPLICATE VIALS

18. Application: Scaling factors  Possible to correlate the amount reconstituted in 30 seconds to that reconstituted in 24 hours with scaling factors  Can be predicted to a precision of ~3% for the commonly encountered extraction solvents Leaf Extraction SolventScaling FactorCV (%) Ethyl Acetate (Mixed)5.778.69 Ethyl Acetate (Static)2.359.40 Hexane-Ethyl Acetate (95:5, v/v)3.443.26 Hexane2.433.14 Ethanol1.463.08

19. Application: Solvent Selectivity  Ethanol showed the highest selectivity for artemisinin and had the greatest extraction extent under the conditions investigated Leaf Extraction SolventExtract Artemisinin PurityTotal Artemisinin Extracted Mean (wt%)CV (%)Mean (mg)CV (%) Ethanol16.910.7410.340.08 Ethyl acetate (no mixing)12.661.867.340.43 95:5 hexane:ethyl acetate11.291.326.900.09 Ethyl acetate (mixing)10.714.249.320.27 Hexane9.821.106.000.07

20. Conclusions  Waxy extract residues provide a barrier that hinders artemisinin re- solubilisation. This has a critical impact on the detected amount of artemisinin and could lead to large underestimates -Literature needs to clarify method and duration of reconstitution for clarity and repeatability -Industry needs set reconstitution procedures  Scaling factors can be used as a method to reduce long reconstitution times  Ethanol was found to be the optimal extraction solvent in terms of total artemisinin extracted and purity of the extract under the conditions investigated

21. Further Work  Investigate more thoroughly the use of ethanol as an extraction solvent and develop methods to purify extracts efficiently  Examine improved methods of overcoming the extended reconstitution times  Examine whether similar effects are observed during column chromatography for purification. Waxy residues are observed to form on the surface of the solid phase

22. Acknowledgements: Dr Chris Preston, Bio Project Consulting Ltd Dr Rachel L Gomes, Department of Chemical and Environmental Engineering, University of Nottingham Afro Alpine Pharma Ltd, Kabale, Uganda More Information: Pilkington, J. L., Preston, C. and Gomes, R. L. (2012) The impact of impurities in various crude A. annua extracts on the analysis of artemisinin by liquid chromatographic methods. Journal of Pharmaceutical and Biomedical Analysis. Article in Press.