Understanding ADT24 for Assessing Thermal Risk in Pharmaceutical Processes

1. When and Why do We Use ADT24to Assess Thermal Risk? Shasha Zhang Process Hazard Evaluation, Chemical Process Development Bristol-Myers Squibb (BMS) P2SAC Fall 2022 Conference Dec 14th-15th2022 Highly Confidential

2. Brief Introduction of BMS PHE Group Where we are Process Hazard Evaluation Drug Substance Operations and Supply CPD What we do Guidance on safety program at partners and support for tech transfer Dust and other testing associated with chemical hazards Hazard evaluation and testing for chemical process Oversight of process safety testing for any production > 20 L Who we support CPD EM CSO Vendors Discovery EHS Lab & KL External Manufacturing Clinical Supply Operation Division/Therapeutic Area 2 Highly Confidential

3. Stoessel Criticality Stoessel Criticality is widely used by many pharmaceutical companies to evaluate thermal risk. ADT24, also known as TMR24and Td24, is the temperature at which decomposition reaches its maximum rate in 24 hours under adiabatic condition Stoessel Diagram Temperature ADT24 Tb ADT24 ADT24 Tb Tb MTSR MTSR MTSR MTSR MTSR ADT24 ADT24 Tb Tb Tp 1 2 4 5 3 Criticallity Tad:Adiabatic Temperature Rise MTSR : Maximum Temp Synthesis Reaction – usually Tp+ Tad Acceptable risk NOT allowed Division/Therapeutic Area 3 Highly Confidential

4. When is ADT24needed to address thermal risk? Tad:Adiabatic Temperature Rise TMR: Time to Maximum Rate ADT24/TMR24is the temperature at which decomposition reaches its maximum rate in 24 hours under adiabatic condition Risk = Probability x Severity Severity – Screening will provide some ideas High Medium 50 oC < TadDecomp <200 oC Low Probability – Based on decomposition kinetics Low TMR > 24 hrs Medium 8 hr > TMR < 24 hr High TMR < 8 hr ADT24/TMR24as a probability indicator is triggered if the severity is medium to high TadDecomp > 200 oC   TadDecomp < 50 oC     Division/Therapeutic Area 4 Highly Confidential

5. When is ADT24needed to address thermal risk? Practical procedure Risk = Probability x Severity Severity – Screening will provide some ideas High Medium 50 oC < TadDecomp <200 oC Low Probability – Based on decomposition kinetics Low TMR > 24 hrs Medium 8 hr > TMR < 24 hr High TMR < 8 hr ADT24/TMR24as a probability indicator is triggered if the severity is medium to high Determine energy potential TadDecomp > 200 oC   TadDecomp < 50 oC  < 100J/g >100J/g Energy from DSC? Tad > 50°C    Assess probability of triggering Decomposition (ADT24) Process presents no thermal risks No need to waste resources! Assess Criticality by comparing MTSR and ADT24 Division/Therapeutic Area 5 Highly Confidential

6. When to Study ADT24at BMS? • Temp <=200 °C, heat > 200 J/g (Tad of 100 °C) ADT24 • Temp <=200 °C, heat < 100 J/g (Tad of 50 °C) No need Onset =< 200 °C Onset > 200 °C Energetic /Explosive? 100 J/g <= Heat <= 200 J/g Kinetics / heat flow? Heat > 200 J/g Heat < 100J/g Yes No Slow Fast Further study Low risk ADT24 Simple decision tree based on DSC data if no significant gas evolution Division/Therapeutic Area 6 Highly Confidential

7. For some marginal cases, thermogram needs to be reviewed to evaluate the necessity Is the Study of ADT24 Needed? BMS coupound 20 wt% DBDMH in DMF • Energy: 107 J/g; Onset 124 °C • Energy: 157 J/g; onset: 64°C • Vendor suggested a ADT24of 40 °C. • No need to study ADT24 • ADT24is determined to be 34°C. 7 Division/Therapeutic Area Highly Confidential

8. How to Determine ADT24 Kinetic approaches Isoconversion method: DSC/AKTS (Advanced Thermal Analysis Software) Run DSCs Estimate Ea and A Used by BMS  dt   d E  (     ) A exp f RT  TMRad (ΔHr=-406.4Jg^-1, Cp=2Jg^-1K^-1, Phi=1, ΔTad=203.2°C) Predict time to TInitial=102.5°C maximum rate 300 R = -0.99957 Heat = -406.397 ± 13.027 J/g 0.5 K/min : -417.506 J/g (Peak Height < 1 mW !) 1 K/min : -414.793 J/g (Peak Height < 1 mW !) 2 K/min : -405.052 J/g 4 K/min : -409.931 J/g 8.02 K/min : -384.705 J/g 0.003 250 Temperature (°C) Reaction rate (-/s) 0.002 200 150 0.001 TInitial=96.5°C TInitial=90.9°C 100 TInitial=82.5°C 0 100 120 140 160 180 200 220 0 5 10 15 20 25 Temperature (°C) Reaction rate Time (h) TMRad (DeltaHr=-406.4Jg^-1, Cp=2Jg^-1K^-1, Phi=1, DeltaTad=203.2°C) Adiabatic calorimetry Directly measure the kinetic curve  Extrapolate the beginning of the curve 500 600 400 Temperature, oC 400 Pressure, psi 300 200 200 0 100 -200 Division/Therapeutic Area 500 8 Highly Confidential 550 600 650 700 Time, min

9. How to Determine ADT24 Estimation approaches Traditional ‘safety Margin’ approach - ‘Distance rule’; – Tsafe= Tonset– X oC Equation-based estimation – Use onset temperature and instrument detection limit assuming Ea= 50 kJ/mol & zero-order kinetics – Empirical equation for DSC: Onset temperature as input • Usually overly conservative •  DEKRA/Chilworth Employed by some vendors •  Used by Legacy Celgene, but labelled as preliminary ADT24 • • NOT BMS preferred approaches • May cause confusion Stoessel, F., Thermal Safety of Chemical Processes Division/Therapeutic Area 9 Highly Confidential

10. Example 1: Production of an Early Intermediate • • • A production of 250 kg of intermediate was executed at vendor’s plant. Desired reaction: RC1 data shows an adiabatic temperature rise of 41 °C. DSC on distillation stream shows an exotherm of -312J/g from 131 °C. ADT24is 43 °C based on equation-based estimation by a vendor. High risk MAT =MTSR = Tp + Tad = 67 °C > ADT24Criticality 5 Tb = 64.7°C Distillation Temp of 45 °C > ADT24 MAT (Maximum Achievable Temperature) MTSR (Maximum Temperature of Synthetic Reaction) Division/Therapeutic Area Highly Confidential Data collected by BMS Partner

11. Example 1: Internal DSC/AKTS Study Acceptable risk • A better method is needed to quantify ADT24. DSC on product shows an exotherm of -410 J/g with an onset of 116 °C. • MAT =MTSR = 67 °C. ADT24is determined as 82 °C Tb = 64.7°C Tb< MTSR < ADT24Criticality 3 Distillation Temp of 45 °C < ADT24 • DSC/AKTS analysis shows ADT24is 82 °C. Product is not isolated so ADT24is considered conservative • R = -0.99957 Heat = -406.397 ± 13.027 J/g 0.5 K/min : -417.506 J/g (Peak Height < 1 mW !) 1 K/min : -414.793 J/g (Peak Height < 1 mW !) 2 K/min : -405.052 J/g 4 K/min : -409.931 J/g 8.02 K/min : -384.705 J/g Safety diagram (ΔHr=-406.4±13J/g, Cp=2J/g/K, Phi=1) 120 0.003 115 TInitial-1°C, ΔTad=196.7°C 110 Time : 2 hours Temp : 102.5 °C Reaction rate (-/s) 105 Temperature (°C) 0.002 Time : 4 hours Temp : 96.5 °C 100 95 Time : 8 hours Temp : 90.9 °C 0.001 90 TInitial, ΔTad=203.2°C Time : 24 hours Temp : 82.5 °C TInitial+1°C, ΔTad=209.7°C 85 X : 24 Y : 82.5 80 0 0 5 10 15 20 25 30 35 40 Time (h) Safety diagram (DeltaHr=-406.4±13J/g, Cp=2J/g/K, Phi=1) 100 120 140 160 180 200 220 Division/Therapeutic Area 11 Temperature (°C) Highly Confidential Reaction rate

12. Example 1: Risk Mitigation Overview • An overly conservative estimate of ADT24may greatly affect assessment on safety. • The ADT24provided by vendors may lead to confusion – it is highly recommended to contact safety specialist for clarification. High risk Acceptable risk Importance of ADT24study MAT =MTSR = 67 °C. ADT24is determined as 82 °C Tb = 64.7°C Tb< MTSR < ADT24Criticality 3 Distillation Temp of 45 °C < ADT24 MAT =MTSR = 67 °C ADT24is estimated to be 43 °C. MTSR > ADT24Criticality 5 Distillation Temp of 45 °C > ADT24 Division/Therapeutic Area 12 Highly Confidential

13. Example 2: Thermal Stability of Rxn Mixture An early intermediate was produced on a commercial scale. The condensation is endothermic with a Tad of -12 °C. C80 data of reaction mixture shows an exotherm with an enthalpy of -420 J/g and an onset of 135 °C. Other ARSST and ARC tests confirmed that the exotherm poses a safety concern. • • • • Heating rate: 0.5 K/min Sample mass: 479mg Division/Therapeutic Area Highly Confidential

14. Example 2: Summary of ADT24study Sample used 10 mL Phi Onset Temp 140 °C ADT8 ADT24 ARSST ~1.05 124 115 C80+AKTS 2 mL - 136 °C 118 104 Phi Tec II 50 mL 1.14 135 °C 118 109 ARC (BMS) 5 mL 1.27 123 °C 117 110 ARC (vendor) 3 mL 1.91 131 °C 115 104 If ’50 K safety margin’ or ‘equation-based estimation’ approach is used, the process temp will probably be set around 76 °C, which is too low for the reaction to run. Consistent results of ADT24were obtained from different kinetic methods. Combining with reaction calorimetry data, the criticality index of the process is rated as “2” in Stoessel system. The process was safely implemented at commercial scale. • • • • Division/Therapeutic Area Highly Confidential

15. Example 3: ADT24Study of Tetrahydroxydiboron Tetrahydroxydiboron has been used in many BMS projects. • ADT24study is important due to its fast decomposition rate and large energy (~1600J/g) Onset temperature is in the range of 95 -125 ºC depending on experimental condition. • • ADT24is in a narrower range of 70-82 ºC, indicating ADT24is a more reliable parameter to set the maximum operation temperature limit. • ADT24, ◦C 75 72 76 77 71 82 70 70 Exp # 1 2 3 4 5 6 7 8 Test Method DSC/AKTS DARC DARC DARC DARC DARC ARC ARC Sample load, g mg 0.199 0.383 0.751 1.02 0.397 1.44 0.98 Onset, ◦C >100 125 110 115 95 120 110 110 • Estimation of ADT24using onset temperature may introduce larger variability. Division/Therapeutic Area 15 Highly Confidential DARC data collected by D. Vanyo

16. Summary When to study ADT24? • ADT24as a probability indicator is triggered if the severity of decomposition is medium to high. − Simplified answer: an exotherm energy > 100 J/g below 200 °C. Why to use ADT24? • Using an overly conservative estimate of ADT24may greatly affect the assessment on safety. Simple screening test with distant rule or equation-based estimation may be used by BMS vendors. • A real kinetic study of ADT24delivers reasonable safety margin. − Kinetic approaches are preferred by BMS − Kinetic approaches provide realistic ADT24, and have been demonstrated by many portfolio projects at BMS. • ADT24is a more reliable parameter to set maximum operation temperature. Estimation of ADT24using onset temperature may introduce larger variability. − − Division/Therapeutic Area 16 Highly Confidential

17. Acknowledgement • Simon Leung, Tao Chen, Douglas Mcleod, Mark Lindrud, Junying Fan, Buck Holloway, Melanie Miller, Paul Fernandez, Nicolas Cuniere, Fransisco Gonzalez Bobes, Chris Sfouggatakis, Dimitri Skliar, Subha Mukherjee, Bahar Inankur, Patrick Sipple, Ketleine Georges, Chris Tickner, Dale Vanyo, Alan Fritz, Steven Chan • BMS partner’s safety group and project team for safety study. Division/Therapeutic Area 17 Highly Confidential