Variations in internal standard response Some thoughts and real-life cases Olivier Le Blaye Inspector 21 November 2013 EBF 6th Open Meeting Disclaimer The views expressed in this presentation are mine and may not represent those of either ANSM, EMA, the PKWP or the GCP IWG Agence nationale de sécurité du médicament et des produits de santé 1 Preliminary observation: choice of IS A minority of the studies I inspect use a stable-isotope labelled IS (SIL IS) Very few have a structural analogue Most IS used are not structurally or chemically related to the analyte Some have the same indication: so what ? Agence nationale de sécurité du médicament et des produits de santé 2 WTF ? Analyte: ethinyl estradiol Derivatisation with dansyl chloride IS: diltiazem No derivatisation Agence nationale de sécurité du médicament et des produits de santé 3 Variations in internal standard response Different situations Isolated variation (limited number of isolated samples) Systematic differences, trends… Agence nationale de sécurité du médicament et des produits de santé 4 Variations in internal standard response Isolated variation Easy to detect and to handle Often seen in SOPs: repeat if IS response deviates by more than x % from mean (of CC / QC ? of whole run ?) Influence on overall study results limited, unless Cmax sample Agence nationale de sécurité du médicament et des produits de santé 5 IS variation: individual samples 50000 Double spike ? Sample processing problem ? 40000 30000 CC QC Subject 20000 10000 Sample processing problem ? Air bubble ? 0 0 10 20 30 40 50 60 70 Agence nationale de sécurité du médicament et des produits de santé 6 IS variation: individual samples Usually well described in SOPs, monitored by labs, few inspection findings If variation due to IS addition error: stable isotope labelled IS will not compensate If variation due to other reasons: SIL IS may compensate. But: Low response: analyte S/N ? LLOQ ? High response: linearity ? Agence nationale de sécurité du médicament et des produits de santé 7 Variations in internal standard response Systematic differences, trends… Multiple possible reasons May or may not affect the results More difficult to capture in SOP, may require lab investigations, scientific approach Agence nationale de sécurité du médicament et des produits de santé 8 Variations in internal standard response Millions Systematic differences 3,00 2,50 2,00 1,50 1,00 0,50 0,00 0 Standard QC S14 S15 S16 Agence nationale de sécurité du médicament et des produits de santé 160 9 Systematic differences SOP with ± 50 % rule may not help ! ± 50 % of mean IS response of CC / QC samples 50000 40000 30000 20000 10000 0 1 Standard QC S10 S11 Agence nationale de sécurité du médicament et des produits de santé 118 10 Systematic differences SOP with ± 50 % rule may not help ! ± 50 % of mean IS response of all samples 50000 40000 30000 20000 10000 0 1 Standard QC S10 S11 Agence nationale de sécurité du médicament et des produits de santé 118 11 Variations in internal standard response Systematic differences Applying blindly an SOP with ± x % acceptance criteria may result in the blind acceptance of possibly inaccurate results ! Agence nationale de sécurité du médicament et des produits de santé 12 Variations in internal standard response Trends – case 1 100000 80000 60000 CC QC Subject 40000 Cmax ISR: fails -70 % 20000 0 0 10 20 30 40 50 60 70 Agence nationale de sécurité du médicament et des produits de santé 13 Variations in internal standard response Trends – case 1 Lab did not monitor IS response SOP with ± x % rule may not help anyway IS response obviously results in inaccurate results Tmax at 10’ with bolus IV administration; other datasets: 1-2’ ISR: 70 % difference What % difference in IS would mean inaccurate results ? Agence nationale de sécurité du médicament et des produits de santé 14 Variations in internal standard response Trends – case 2 400000 350000 300000 CC QC Subject 250000 200000 150000 0 10 20 30 40 50 60 Agence nationale de sécurité du médicament et des produits de santé 15 Variations in internal standard response Trends – case 2 Gradual increase in subject IS, decrease in QC IS Are QCs representative of subject samples ? Accuracy of subject samples ? Subject samples with highest IS result were re- analysed No IS response trend in repeat run Repeat result confirmed initial value, data accepted Agence nationale de sécurité du médicament et des produits de santé 16 Variations in internal standard response Trends – case 3 30000 25000 20000 CC QC Subject 15000 10000 5000 0 0 10 20 30 40 50 60 70 Agence nationale de sécurité du médicament et des produits de santé 17 Variations in internal standard response Trends – case 3 IS response systematically lower every 11 and 12 samples Post-inspection root-cause analysis IS added with Gilson 401 dilutor, series of 12 fractions Dilutor checked periodically, but only 10 fractions Due to piston seal problem: fractions 11 and 12 too low Incorrect IS addition resulted in inaccurate results Affected all runs, possibly other studies Agence nationale de sécurité du médicament et des produits de santé 18 Variations in internal standard response Trends – case 4, ISR run 90000 80000 70000 CC QC Subject 60000 50000 40000 30000 0 10 20 30 40 50 60 70 Agence nationale de sécurité du médicament et des produits de santé 19 Variations in internal standard response Trends – case 4, ISR run 90000 80000 70000 60000 CC QC Subject 3 ISR results fail, result higher than initial 50000 4 QCs fail, result too high. IS too low ? 40000 30000 0 10 20 30 40 50 60 70 Agence nationale de sécurité du médicament et des produits de santé 20 Variations in internal standard response Trends – case 4, ISR run Low IS response could be the reason for QC and ISR failure in this run Variations in IS response appeared to be the reason for failure of several run / extraction batches in this study – but not always: some runs with low or high IS response in QC sets, result OK IS not monitored during study, issue not identified, not investigated Agence nationale de sécurité du médicament et des produits de santé 21 Variations in internal standard response Trends – case 5 220000 Extraction batch 1 200000 180000 CC QC Subject 160000 140000 120000 Extraction batch 2 100000 0 10 20 30 40 50 60 70 Agence nationale de sécurité du médicament et des produits de santé 22 Variations in internal standard response Trends – case 5 Samples processed in 2 batches per run, injected in the processing order CC, 2 sets QC, period 1 2 sets QC, period 2 from the same subject Difference in IS response between the two extraction batches EMA guideline: acceptance criteria for QC samples First for the whole run Then for each extraction batch Agence nationale de sécurité du médicament et des produits de santé 23 Variations in internal standard response Systematic differences, trends Multiple possible situations Multiple possible reasons Some may affect accuracy, others not Difficult to plan / describe in SOP Criteria such as ± x % may not be relevant May not be sensitive enough: ± x % inaccuracy ? If limits too tight: may result in numerous unneeded repeats ISR results can help in discussion (too late ?) Agence nationale de sécurité du médicament et des produits de santé 24 Variations in internal standard response Systematic differences, trends “We are using an SIL IS anyway, gold standard, will compensate for any variation, no concern, no need to check IS response” Agence nationale de sécurité du médicament et des produits de santé 25 Variations in internal standard response Systematic differences, trends “We are using an SIL IS anyway, gold standard, will compensate for any variation, no concern, no need to check IS response” Not an acceptable answer if the root cause for the variation has not been identified SIL IS can compensate for some sources of variations, but not all ! If you don’t know why the IS varies, you don’t know whether the SIL IS will compensate Agence nationale de sécurité du médicament et des produits de santé 26 Variations in internal standard response Systematic differences, trends IS variations may trigger laboratory investigations to identify root cause Depending on the result: may require the re-analysis of study samples We, inspectors, should be ready to accept decisions which are science-driven and not just based on SOPs, if: Based on facts and solid scientific arguments Well documented Lack of SOP or inadequate criteria are not valid reasons to accept obviously inaccurate results Agence nationale de sécurité du médicament et des produits de santé 27 See Tan et al., Journal of Chromatography B, 877 (2009) 3201-3209 for 12 more cases ! Thank you for your attention ! Agence nationale de sécurité du médicament et des produits de santé 28 Avertissement • Lien d’intérêt : personnel salarié de l’ANSM (opérateur de l’Etat). • La présente intervention s’inscrit dans un strict respect d’indépendance et d’impartialité de l’ANSM vis-à-vis des autres intervenants. • Toute utilisation du matériel présenté, doit être soumise à l'approbation préalable de l’ANSM. Warning • Link of interest: employee of ANSM (State operator). • This speech is made under strict compliance with the independence and impartiality of ANSM as regards other speakers. • Any further use of this material must be submitted to ANSM prior approval.