Organic Process Research & Development Article pubs.acs.org/OPRD Quality by Design in Action 2: Controlling Critical Material Attributes during the Synthesis of an Active Pharmaceutical Ingredient Abdul Qayum Mohammed/'* Phani Kiran Sunkari,* Amjad Basha Mohammed,§ P. Srinivas,*'* and Amrendra Kumar Roy*'* ^CTOTII, Dr. Reddy's Laboratories Ltd, Plot 116, 126C and Survey number 157, S.V. Co-operative Industrial Estate, IDA Bollaram, Jinnaram Mandal, Medak District, Telangana 502325, India ^Department of Chemistry, Osmania University, Hyderabad, Telangana 500007, India ^Research and Development, Integrated Product Development, Innovation Plaza, Dr. Reddy's Laboratories Ltd, Bachupally, Qutubullapur Mandal, Rangareddy District, Telangana 500072, India ABSTRACT: Quality by Design (QbD) is of paramount importance not only for patient safety but also for the timely and uninterrupted supply of products at affordable prices into the market. Both of these objectives can be achieved only through a robust process, and one of the major obstacles for developing a robust process is the quality of input materials and reagents used for the manufacture of active pharmaceutical ingredients (APIs). This article demonstrates the use of QbD methodology to optimize the quality of input materials and make the process more consistent, thereby reducing the variation in the quality of API produced. This article highlights the use of failure mode and effect analysis (FMEA) for the unbiased identification of critical process parameters and critical material attributes associated with the manufacturing of key starting materials, which are later used as input for the design of experiments (DoE) study that is used for the optimization. ■ INTRODUCTION The main aim of any Quality by Design (QbD) process is to address the variability in the critical quality attributes (CQAs) of an active pharmaceutical ingredient (API) to ensure that the risk to patients' health is mitigated. QbD also helps in controlling the cost of medicines and ensuring uninterrupted supply of medicines into the market. There are many sources of variability, and one of the major sources is the inconsistent quality of key starting materials (KSMs) and reagents used in the production process. Failure to study and properly control the quality of the KSMs can have far-reaching consequences for not only the process robustness but also the business, as shown in Table 1. Table 1. Effect of process inconsistency from the supplier and/or manufacturer on API quality manufacturer's API process robust not robust supplier's KSM process robust case-1: robust process case 2: variability due to process not robust case 3: variability due to KSM case 4: disaster From case 1 in Table 1, it is evident that consistency in the CQAs of an API is possible only if both the manufacturer and the supplier have robust processes for the API and KSM, respectively. Any kind of reprocessing/rework of an unsuitable KSM at the manufacturer's end is not a viable option, as it would increase the cost of production, which has to be borne either by the manufacturer or the patients. Hence, it is important for a manufacturer to engage the suppliers in its QbD journey in order to eliminate at least one source of variation (i.e., from KSM) from the manufacturing process. Another analogous scenario is the multistep synthesis, where the quality of the penultimate stage (KSM manufactured in-house) becomes detrimental to the CQA of the final API. In QbD terms, the desired quality of the KSM is described as critical material attribute (CMA). This article demonstrates the use of QbD to optimize the reaction parameters in order to achieve the desired quality of the KSM (the penultimate stage), which in turn results in minimizing the variability at the API stage. In this regard, we have reported in a companion article1 a possible sequence of steps involved in the implementation of QbD and illustrated it with a case study, where the effects of critical process parameters for stage 5 (CPPS) and critical material attributes for stage 5 (CMAs)a on the CQAs of the final API (compound 5, Scheme l) were studied. The present article is an extension of the companion article in which QbD is used in a similar way to control the CMAS in order to have a robust process at the API stage, as shown in Figure 1 and Scheme 1. The various terminologies used in the present article are explained for the clarity of readers. As shown in Figure 1, the CQAs, CPPs, and CMAs associated with the final API (stage 5) are denoted as CQAS, CPPS, and CMAS, respectively. CMAS itself is affected by two things: the critical process parameters related to stage 4, denoted as CPP4, and the critical material Special Issue: Application of Design of Experiments to Process Development Received: September i5, 20i4 Published: January 2i, 20i5 ACS Publications e 2015 American Chemical Society 1645 DOI: 10.1021/op500297g Org. Process Res. Dev. 2015,19,1645-1654 Organic Process Research & Development Article Scheme 1. Synthetic route to API hydrochloride and impurities observed at the final stage" (CH2)n ,N_ k + R( R2 NH3CI 5 API Hydrochloride N- R2 ,0 0 /^(CH2)n R2 (H2C)n- HN—\—mh Hydrolyzed impurity Rj _J> R2 > ,0 0 ^(CH2)n HN—^ ^NH O Dimer impurity 7 (CH2)n HN- Laotam impurity 8 "Reagents: (a) SOClj, toluene; (b) potassium phthalimide, DMF/H20; (c) 40% aqueous methylamine solution; (d) EtOAc/HCl gas. This part of the work reported earlier Specification of compound 3 Specification of Methylamine solution Specification of Final API (compound 5) CPP4For Stage 4 Presentwork describesthe use of QbD in optimizing the effect of CPP4 & CMA, on CMA, CQA of API Figure 1. Various abbreviations used in the present article. Subscripts represent stage numbers. Table 2. Screening of MA5 specifications is it a CMA5? remarks A compound 4 1 assay as per analysis no compound 4 is added to the next stage based on the assay of compound 4 in the crude reaction mass. 2 residual Toluene no B impurities 1 unreacted (3) NMT 1% yes it was desired to keep these impurities at minimum level in order to have optimum yield. 2 hydrolyzed Imp. (V) NMT 3% yes 3 dimer Imp. (8) NMT 3% yes 4 yield > 80% yes it was desired to have > 80% yield for optimum RMC. C EtOAc/HCl HCI concentration NLT 8% 8-12% no HCI concentration to be in range of 8-12%. 1646 DOI:10.102Vop500297g Org. Process Res. Dev. 2015,19,1645-1654 Organic Process Research & Development Article Scheme 2. Synthetic scheme for stage 4 °Y>H2)n R'NxR2 ^NH2 Table 3. CMA4 for stage 4 MAs no. raw material purity 1 compound 3 NLT 98% is it a assay range CMA4? remarks >98% yes starting material 2 methylamine 40% aqueous 35—40% yes reagent for solution reaction Half-Normal Plot 0-00 006 0.13 0.19 I Standardized Effect| Pareto Chart 026 O 14.71 n n Figure 2. Half-normal plot and Pareto chart for unreacted 3. attributes of compound 3 and methylamine solution, which are together denoted as CMA4. In the companion article,1 the focus of the QbD was to identify and optimize the important process parameters (CPPS) along with important material attributes of the input materials ft Reaction Temperature Figure 3. Effect of CPP4 on unreacted 3 after 5 h. Half-Normal Plot 0.6S I Standardized Effect| Pareto Chart Figure 4. Half-normal plot and Pareto chart for impurity 6. (compound 4 and EtOAc/HCl solution), which together constitute CMAS. The present article deals with the optimization of CMAS (i.e., the quality of compound 4) by controlling the CPP4 and CMA4 involved in the deprotection of compound 3 to give compound 4. 1647 DOI: 10.1021/op500297g Org. Process Res. Dev. 2015,19,1645-1654 Organic Process Research & Development Article Table 4. FMEA-2 for the identification of CPPs for stage 4 effect on CQAs of stage 5 potential effect(s) of failure on s. N 0 unit operations or process parameters (PPs) potential failure mode yield of 4 Purity -3 V 5 « i-c hydrolyzed impurity 6 dimer impurity 7 stage (4) API (5) failure mode present control occurrence severity _o S V -= s U _s RPN proposed control remarks more quantity of toluene t $ $ t t no impact between 9-12 volumes no effect error in manual charging charging by 5 2 2 20 not required keep it constant charge toluene 11 volumes less quantity of toluene t $ t t t flow meter 5 2 2 20 not required between 9-12 volumes 1 Into the reactor at 30±5°C charging of toluene at high temperature t t t t t no Impact no effect failure of steam inlet valve ensure steam valve is closed 5 7 2 70 replace steam line with hot water line charging temperature to be held constant at 30±5°C charging of toluene at low temperature t t $ t 25- 40 °C temperature fluctuation in RT water no action 5 2 5 50 not required 2 charge of compound 3 into the reactor at 30±5°C more quantity of compound 3 ■i t t t incomplete reaction unreacted compound 3 carried to API stage weighing error error in methyl amine assay escape of methyl amine from container calibration of weighing balance on daily basis in ware house qualifying methyl amine based on vendor COA 2 7 5 70 restricting the batch size in multiples of 50 kg suppliers to give compound 3 in 50 kg bags reanalysis of methylamine just before use batch size to be held constant less quantity of compound 3 $ t t t less yield no impact weighing error error in methyl amine assay 3 7 5 105 Suppliers to give compound 4 in 50 kg bags for charging of compound 3 at high temperature $ t t t $ no impact between 25- 40 °C no impact failure of steam inlet valve ensure steam valve is closed 5 7 3 105 replace steam line with hot water line charging temperature to be held constant at 30±5°C charging of compound 3 at low temperature t t t t no impact between 25- 40 °C no impact temperature fluctuation in RT water no action 5 2 3 30 not required 3 stir the reaction mass for 10-15 minutes at 30±5°C stirring of reaction mass more than required time t z t t t no Impact no Impact manual error no action 3 1 3 9 not required stirring time to be held constant between 15-20 minutes heating of reaction mass more than required t $ t t t 5 3 3 45 to be held constant between 55±5°C and heating time to be constant between 30-45 minutes 4 heat the reaction mass to 55±5°C heating of reaction mass less than required t t t t no impact as methyl amine is not added no impact manual error ensuring RT water in condenser to stop toluene loss 2 1 3 6 replace steam line with hot water line slow heating of reaction mass $ t t t t 5 1 3 15 fast heating of reaction mass $ t t t t 5 1 3 15 1648 DOI: 10.1021/op500297g Org. Process Res. Dev. 2015,19,1645-1654 Organic Process Research & Development Article Table 4. continued unit operations or process parameters (PPs) potential failure mode effect on CQAs of stage 5 potential effect(s) of failure on stage (4) API (5) failure mode present control proposed control addition of 40% methyl amine solution (CPP-2) low concentration of methyl amine incomplete reaction may give rise to SMUI error in methyl amine assay escape of methyl amine from container qualifying methyl amine based on vendor COA cap sealed after sampling reanalysis of methylamine just before use specification of assay to be constant between 35-40% high concentration of methyl amine maximum available concentration is 40% no impact none, as assay cannot be more than 4D% QC analysis not required more eq, of methyl amine to be investigated less eq. of methyl amine manual error give rise to impurities with less yield issue only required no. of carboys impact to be studied using DoE 225 add methyl amine solution at 55±5°C (CPP-1) addition at high temperature addition at low temperature to be investigated, as it can escape before it reacts manual error use of steam and temperature indicator 243 120 replace steam line with hot water line impact need to be studied by DoE along with reaction temperature more maintenance time than the required time to be investigated, as it can escape before it reacts manual error 200 maintain the reaction mass at 55±5°C for 5 Hrs (CPP-3) less maintenance time than the required time log book for recording time impact need to be studied by DoE to be investigated. May give rise to SMUI manual error maintenance at high temperature id be investigated, as it can escape before it reacts maintenance at low temperature incomplete reaction may give rise to SMUI manual error hourly record of temperature and adjusting the temperature accordingly 280 impact need to be studied by DoE separate the organic laver less settling time yield loss less yield manual error settling time of 15 minutes settling time to be 30 minutes concentrate the organic layer to remove toluene residual toluene improper assay calculate the yield of 5 residual toluene giving wrong weight residual toluene not included while reporting failure of hot water and vacuum pump IPC for residual toluene and correction factor to be included while reporting yield increase in desired CQA Good decrease in undesired CQA Good increase in undesired CQA decrease in desired CQA Bad t no effect ofCPPsonCQA yield reporting after OVI corrections 1649 DOI: 10.1021/op500297g Org. Process Res. Dev. 2015,19,1645-1654 Organic Process Research & Development Article Table 5. Summary of FMEA output (CPP4) from Table 4 S. no. unit operations or process parameters (PPs) RPN is it critical? control strategy 1 charge 10 volumes of toluene into the reactor at 30 ± 5 °C <45 no 9—12 volumes 2 charge compound 3 into the reactor at 30 ± 5 °C <45 no 30 ± 5 °C 3 stir the reaction mass for 10—15 min at 30 ± 5 °C 9 no 15 min 4 heat the reaction mass to 55 ± 5 °C <45 no 55 ± 5 °C 5 add methylamine solution at 55 + 5 °C (CPP4-l) 120-243 yes to be tested 6 amount of 40% methylamine solution (CPP4-2) 225 yes to be tested 7 maintain the reaction mass at 55 + 5 °C for 5 h (CPP4-3) 200-280 yes to be tested 8 separate the organic layer in 15 min 45 no 30 min 9 concentrate the organic layer to remove toluene 75 no OVI correction to be given 10 calculate the yield of 5 in the crude reaction mass 75 no Table 6. Ranges for the three CPP4 considered for DoE symbol CPP+ variable unit low (—) high (+) A CPP+-1 reaction temperature °C 50 70 B CPP+-2 amount of methylamine equiv 7 13 C CPP+-3 reaction time h 4 6 ■ APPLICATION OF QBD TO CONTROL THE CMA5 The stepwise QbD process described in the companion article1 was adopted to identify the CPP4 and CMA4 required for controlling all CMAS. Step 1: Listing of All Material Attributes (MA5) of Compound 4 Involved in the Synthesis of the Final API. The maximum number of CQAs pertaining to the final API (5) originated from compound 4. Hence, all of the CQAs (unreacted 3, residual toluene, impurities 6 and 7) of the API stage become the MAS that need to be controlled by optimization of the conversion of compound 3 to compound 4, as shown in Table 2. In addition, the quality of EtOAc/HCl used at stage 5 is also included in MAS. Step 2: Risk Assessment 1: Identifying the CMA5. All of the MAS of in situ-manufactured compound 4 are captured in Table 2, and few of them are identified as CMAS on the basis of criticality. Step 3: Identification of CMA4 and CPP4 Required for the Synthesis of Compound 4. After the CMAS associated with compound 4 were identified, it was important to identify the CMA4 (i.e., the quality of compound 3 and of methylamine) and CPP4 that are critical to obtain the desired CMAS. Step 3.1: Identification 0fCMA4. The main inputs involved in the manufacturing of compound 4 are compound 3 and methylamine solution (Scheme 2). Hence, the material attributes of both of the inputs material that are critical to the quality of compound 4 are described as CMA4 and are captured in Table 3. Step 3.2: FMEA-2 for the Identification of CPPs. After defining CMA4 that were affecting CMAS, it was then time to identify the CPP4 that were critical to CMAS. As described before, a risk-based analysis of the process was used for the identification of CPP4, and this risk assessment was done using failure mode and effect analysis (FMEA). However, before FMEA is started on any process, it is important to have a process description, as it is the main input for the FMEA. The process involved in the manufacture of compound 4 is briefly described below: Toluene and compound 3 are charged into an round-bottom flask, and the mixture is stirred for 10—15 min and then heated to 55 ± 5 °C. Then 40% aqueous methylamine solution is added at 55 ± 5 °C, and the resulting mixture is further maintained at 55 ± 5 °C for 4—6 h for completion of the reaction. The reaction mass is then cooled to 50 ± 2 °C, followed by separation of the toluene layer. The aqueous layer is once again extracted with toluene, and the combined toluene layers containing the free-base API 4 are concentrated under vacuum below 50 °C. After the entire toluene layer is distilled, the reaction mass is cooled to 30 ± 5 °C and sent for assay analysis. On the basis of the assay, this crude mass is then directly taken for the final stage, where it is converted to its hydrochloride form (5)." Each unit operation described above was subjected to an extensive FMEA procedure by a cross-functional team (R&D, AR&D, PE, and Production), as captured in Table 4. This FMEA helped in filtering out the three CPP4 (reaction time, reaction temperature, and amount of methylamine) on the basis of high risk priority numbers (RPNs), which were then taken as the main output of any FMEA procedure. As summarized in Table 5, there were three CPP4 that were to be studied for their impact on the CMAS of compound 4, and the remaining seven PPs were held constant. Apart from this, Table 7. Results of the 23 full factorial design factors responses (CMA5 from Table 2) CPP4-1: reaction temperature (°c) CPP4-2: amount of methylamine (equiv) CPP4-3: reaction time (h) unreacted 3 (%) hydrolyzed impurity 6 (%) dimer impurity 7 (%) yield (%) 50 7 4 0.08 2.26 1.49 85.63 70 7 4 0.5 2.69 2.55 75.00 50 13 4 0.01 0.62 0.25 87.20 70 13 4 0.1 1.19 0.56 85.00 50 7 6 0.07 1.32 0.94 83.35 70 7 6 0.52 1.83 1.55 80.37 50 13 6 0.01 0.54 0.13 86.63 70 13 6 0.08 0.60 0.21 79.98 1650 DOI: 10.1021/op500297g Org. Process Res. Dev. 2015,19,1645-1654 Organic Process Research & Development \ \ \ 5.50 — \ \ \ X tion time 5.00 — M.50 pi pol Reac \^ ü 4.50 — 4.00 — [zoo 1 \ 1-\-r- 9.0C 10.0C B: Methyl Amine eq. Figure 5. Effect of CPP4 on hydrolyzed impurity 6 at 60 °C. two CMA4 (Table 3) were also well-defined prior to any further optimization. Step 4. Optimization of the Effect of CMA4 and CPP4 on the CMA5. Step 4.1. Optimization of the CMAs. It is important to control the CMA4 (i.e., the quality of compound 3 and methylamine) in order to have control over CMAS (the desired specifications of compound 4). The CMA4 were already well-defined as shown in Table 3. It was then time to optimize the CPP4 affecting the conversion of compound 3 to compound 4. Step 4.2. Optimization of the Effect ofCPP4 on CMA5. A 23 full factorial experimental design was planned to study the effect of three CPP4 (outcome of FMEA analysis; Tables 4 and 5) on CMAS, keeping all of the other PPs constant at the desired levels (Table 5). The investigational ranges for the three CPP4 considered for the DoE are given in Table 6, and the results of the full factorial design are given in Table 7. The analyses of the DoE results for the various CMAS are discussed in the following sections. 4.2.1. Effect of the Three CPP4 on Unreacted 3. The half-normal plot and the Pareto chart (Figure 2) and the analysis of variance (ANOVA) (Table 8) show that the unreacted starting material 3 in the reaction mass was influenced not only by the reaction temperature and amount of methylamine but also by their interaction effect. Lower reaction temperature and excess methylamine lead to less unreacted 3 and a greater yield of product 4. A higher level of unreacted 3 may be due to the loss of methylamine at higher temperature. The same is depicted in the contour graph given in Figure 3. 4.2.2. Effect ofCPP4 on Hydrolyzed Impurity 6. In this case, the half-normal plot and the Pareto chart (Figure 4) indicate that the amount of hydrolyzed impurity 6 was affected inversely by the amount of methylamine and the reaction time, whereas the reaction temperature did not have any impact on this impurity. The same conclusion can be drawn from ANOVA analysis (Table 9) and the contour graph (Figure 5). In other words, a higher amount of methylamine and higher reaction time favors a reduction of impurity 6. Table 8. ANOVA table for unreacted 3 source sum of squares degrees of freedom mean square F value p value prob > F model 0.31 3 0.10 928.11 <0.0001 significant A (reaction temperature) 0.13 1 0.13 1178.78 <0.0001 B (amount of methylamine) 0.12 1 0.12 1045.44 <0.0001 AB 0.06 1 0.06 560.11 <0.0001 residual 0.00 4 0.00 cor total 0.31 7 Table 9. ANOVA table for hydrolyzed impurity 6 source sum of squares degrees of freedom mean square F value p value prob > F model 4.10 2 2.05 19.04 0.0046 significant B (amount of methylamine) 3.33 1 3.33 31.00 0.0026 C (reaction time) 0.76 1 0.76 7.08 0.0449 residual 0.54 5 0.11 cor total 4.63 7 Table 10. ANOVA table for impurity 7 source sum of squares degrees of freedom mean square F value p value prob > F model 3.62 1 3.62 14.92 0.0083 significant B (amount of methylamine) 3.62 1 3.62 14.92 0.0083 residual 1.45 6 0.24 cor total 5.07 7 Table 11. ANOVA table for the percent yield source sum of squares degrees of freedom mean square F value p value prob > F model 68.86 1 68.86 19.08 0.0047 significant A (reaction temperature) 68.86 1 68.86 19.08 0.0047 residual 21.65 6 3.61 cor total 90.51 7 1651 DOI: 10.1021/op500297g Org. Process Res. Dev. 2015,19,1645-1654 Organic Process Research & Development Article Pareto Chart Half-Normal Plot III I I . ~\-1-1-1-r Rank Half-Normal Plot [Standardized Effect| Figure 6. Pareto chart and half-normal plot for impurity 7. 1.401 11.201 11 001 10.801 10 601 11.00 12.00 B: Methyl Amine eq. Figure 7. Effect of CPP4 on dimer impurity 7 at 60 °C. 4.2.3. Effect ofCPP4 on Dimer Impurity 7. It is evident from the Pareto chart and the half-normal plot (Figure 6) that the amount of impurity 7 was affected inversely by the amount of methylamine, while the other two CPPs had no effect on it. This fact was augmented by the ANOVA analysis (Table 10) and also by the contour plot (Figure 7) 4.2.4. Effect ofCPP4 on the Yield of Compound 4. The half-normal plot and Pareto chart (Figure 8), ANOVA analysis (Table 11), and contour plot (Figure 9) show that the yield had an inverse relationship with the reaction temperature, while the other two CPP4 had no effect. It might be possible that at higher Standardized Effect| Pareto Chart 13 iL 'S 21 D □ _ Figure 8. Half-normal plot and Pareto chart for the percent yield. Yield % GQ 9.00 - 62 00 1 81 OOl 55.00 60.00 65.00 A: Reaction Temperature Figure 9. Effect of CPP4 on the percent yield for a reaction time of 5 h. temperature methylamine could escape from the reaction mass, thereby decreasing the yield and increasing the amount of intermediate hydrolyzed impurity 6. 4.2.5. Summary of the Effects of CPP4 on CMA5. The contributions of all three CPP4 and their interactions to the four CMAS of compound 4 are captured in Figure 10. Step 4.3. Defining the Design Space for Compound 4. Finally, a design space was generated by defining constraints for 1652 DOI: 10.1021/op500297g Org. Process Res. Dev. 2015,19,1645-1654 Organic Process Research & Development Article % contribution of factors on each response 60.00 40.00 J 0.00 -20.00 -40 00 -60.00 -80.00 I A-Reaction Temperature I B-Methyl Amine eq. C- Reaction time (Hrs) I AB I AC BC ABC Responses Figure 10. Contribution of CPP4 on CMAj of compound 4. Table 12. Criteria for defining the design space Overlay Plot name of CPP/CMA A (reaction temperature) B (amount of methyl amine) C (reaction time) unreacted 3 hydrolyzed impurity 6 dimer impurity 7 yield goal CPP+ °C in range equiv in range h target CMA5 or Response % HPLC minimize % HPLC minimize % HPLC minimize % maximize lower limit 50 10 upper limit 70 12 5.5 E < 0.01 0.53 0.13 NLT 82 all three CPP4 and CMA4 involved in the process, as shown in Table 12. It is worth mentioning that the rest of the process parameters that were not critical were held within their ranges as defined in the FMEA (see Tables 4 and 5). On the basis of the constraints defined for CMAS as shown in Table 12, an overlay plot of all the CPP4 was generated (Figure 11), thereby defining a boundary within which CPP4 could be varied with no effect on CMAS. This amicable region, within which the process meets all of the specifications for CMAS, is shown as the yellow region in Figure 11 and is called as proven acceptable range. This amicable range is defined in Table 12. However, the red rectangle inside the yellow region, which is our normal operating range, becomes the desired design space. Step 5. Defining Control Strategies3 for All of the CMAs and CPPs. The control strategies for all of the CMA4 are presented in Table 3, and the control strategies for all critical/ noncritical process parameters were determined after FMEA analysis (Tables 4 and 5). Finally, the control strategies for the three CPP4 were defined after the DoE study and are captured in I Dirr-er impurity (8) CI: 0.000 Unreacted 3: 0.09 Hydrolyzed impurity 6: 0.S Dimer1 0.5 Yield- 82% I Viele ilg.OOC I I Yield 1 82 0001 Unreacted 3: 0.13 Hydrolyzed impurity 6- 1.7 Dimer impurity 7: 1.46 N yield: S4.76 -1-1- 500-0 5500 60.00 65.00 A: Reaction Temperature Figure 11. Design space (red rectangle) defined for the reaction time of 5.5 h. Table 13. These CPP4 and CMA4 would be controlled and monitored closely in the future, during commercialization, using various process analytical tools (PATs) and statistical process control tools.4 Finally, the specification of compound 4 (CMAS) was optimized on the basis of the design space, as captured in Table 14. It is worth mentioning that even though high levels of impurities at stage 4 could be tolerated in the next stage, the QbD helped in optimizing the reaction conditions, resulting in much lower levels of these impurities (compare Tables 2 and 14). Step 6. FMEA-3: Assessing the Risk Mitigation. The last step of the QbD process was to assess the effect of DoE on the 1653 DOI: 10.1021/op500297g Org. Process Res. Dev. 2015,19,1645-1654 Organic Process Research & Development Table 13. Control strategies for the three CPP4 with their revised RPNs after FMEA-3 FMEA-3" s. no. factor acceptable range control strategy O S D RPN 1 reaction temperature 55 + 5 °C ~52—60 °C (cpp+-i) replace steam line with hot water line 3 7 3 63 2 amount of methylamine solution 9.5—11 equiv (CPP+-2) reanalysis of methylamine solution just before use; specification of assay to be fixed between 35 and 40% 3 7 3 63 3 maintain the reaction mass at 55 + 5 °C 5.5—6 h for 5 h (CPP+-3) replace steam line with hot water line 3 7 3 63 "O = occurrence, S = severity, D = (lack of) detection. Table 14. Final specifications for compound 4 specifications (CMA5) maximum tolerable limit process control limit1* is it a CMA5? remarks 1.1 assay as per analysis as per analysis no it is taken to the next stage on the basis of the assay of 4 1.2 residual toluene as per analysis as per analysis no 1.3 unreacted 3 NMT 1% 0.5% yes even though these would not participate in the next stage, it was desired to keep these 1.4 hydrolyzed NMT 3% 1.5% yes at minimum levels impurity 6 1.5 dimer impurity 7 NMT 3% 0.5% yes "These limits were the outcome of the DoE. RPN of each CPP4 by comparing the RPN with the value before DoE (i.e., as determined by FMEA-2). For the three CPP4, these RPNs decreased significantly, as shown by a comparison of the values in Table 13 with those in Table 4. ■ CONCLUSION This article has demonstrated the stepwise methodology of implementing QbD to determine the CMAs for any KSM. The emphasis was on optimizing the CMAs of the KSM to ensure that the quality of the final API stage would become consistent in the future. In addition, this exercise would eliminate at least one source of variation from the process. It is also evident that if a manufacturer is obtaining a KSM from outside/third party, then it is beneficial for the manufacturer to include the supplier in the QbD journey. Furthermore, the case study illustrates how FMEA can be used for the unbiased selection of CPPs and CMAs, which can then be used as an input for DoE studies. Finally, the operating ranges for all of the CPPs were finalized on the basis of the design space obtained after DoE, thereby providing a robust process. ■ AUTHOR INFORMATION Corresponding Authors *Telephone: +919701346355. Fax: + 91 08458 279619. E-mail: amrendrakr(2>drreddys.com (A.K.R). *E-mail: sripabba85(2>yahoo.co.in (P.S). Notes The present article represents the authors' personal views on the subject. The authors declare no competing financial interest. DRL Communication Number IPDO-IPM 00423. ■ ACKNOWLEDGMENTS We thank DRL management for supporting this initiative. ■ ABBREVIATIONS ANOVA analysis of variance API active pharmaceutical ingredient CMA critical material attribute CPP critical process parameter CQA critical quality attribute DoE design of experiments equiv equivalents FMEA failure mode and effect analysis h hours KSM key starting material MA material attribute NLT not less than NMT not more than PP process parameter QbD Quality by Design RPN risk priority number SMUI single major unknown impurity wrt with respect to variance ■ ADDITIONAL NOTE "The desired specifications of compound 4 and EtOAc/HCl are used as inputs for the manufacture of the final API (see Scheme l). ■ REFERENCES (1) Qayum, M. A; Sunkari, P. K.; Srinivas, P.; Roy, A. K. Org. Process Res. Dev. 2015, DOI: 10.1021/op500295a. (2) Note on detectability: The risk number given to detectability actually shows the lack of detectability: a higher number means that the CQA is not detectable by the analytical method. (3) (a) Lobben, P. C; Barlow, E.; Bergum, J. S.; Braem, A; Chang, S. Y.; Gibson, F.; Kopp, N.; Lai, C; LaPorte, T. L.; Leahy, D. K.; Miislehiddinoglu, J.; Quiroz, F.; Skliar, D.; Spangler, L.; Srivastava, S.; Wasser, D.; Wasylyk, J.; Wethman, R; Xu, Z. Org. Process Res. Dev. 2014, DOI: 10.1021/op500126u. (b) Zhou, G.; Moment, A.; Cuff, J.; Schafer, W.; Orella, C; Sirota, E.; Gong, X.; Welch, C. Org. Process Res. Dev. 2014, DOI: 10.1021/op5000978. (4) Mukundam, K.; Varma, R. N. D.; Deshpande, G. R; Dahanukar, V. H.; Roy, A. K. Org. Process Res. Dev. 2013, 17, 1002. 1654 DOI: 10.1021/op500297g Org. Process Res. Dev. 2015,19,1645-1654