Tips and Tricks for SFC Method Development

Transcription

1 Tips and Tricks for SFC Method Development Jeffrey Kiplinger, Paul Lefebvre, Mickey Rego Averica Massachusetts, USA Waters User s Meeting, Prague CZ, December 2015

2 The Good Enough Concept Good enough requires that the standard be much lower than BEST, but offer valuable compensations Flip camera MP3 music file compression Cloud-based applications Kindle books Micro-clinics

3 Good Enough Chromatography Not the same as Fully Optimized Chromatography Screen many conditions for optimal selectivity Do loading studies Appropriate for large-scale separations, esp. as an ongoing production process Averica s business: Contract chromatography services, small molecule pharmaceutical sector, non-gmp, 5 mg to 500 g scale ~45% chiral separations, average sample size 5-20 grams Task: deliver scalable supply of high value compound fast

4 De-Risking Early Drug Development Best Practice List for Pre-GLP Safety Assessment Average Phys. Chem. prop prediction 94% Genotox pilot (e.g. mini-ames) 87% Off-target assays (e.g. CEREP) 87% herg inhibition (e.g. auto patch clamp) 85% Reactive metabolite detection 85% 1-2 week pilot tox in rodent 82% Genotox prediction (e.g. DEREK) 79% General tox & ADME prediction (e.g. MCASE, TOPKAT) 74% Mouse micronucleus (gene tox) 72% Ames (normal Ames) 68% 3-4 day mini-tox in rodent 68% Safety pharmacology core battery 66% Safety pharmacology telemetry 62% 1-2 week pilot tox in large species 60% herg screening (e.g. Rb efflux) 57% Transporter binding/inhibition 57% Percentage of programs requiring specific assay prior to full development - Source: Drug Safety Executive Council (DSEC) survey of 20 Directors of safety assessment at major pharmaceutical companies, 2011

5 Principles Work with a small set of favored CSPs Move to prep columns rapidly Use work arounds that limit method development effort Vary the co-solvent rather than the CSP Take a big valley Overlay peaks in stacking when possible

6 RegisPack AD IA CCA-EX-b12 RegisCell OD CCO-EX-b14 OJ CCJ AS Lux-2 IC Whelk-O1 CC4 CLA-1 Sulfinpyrazone yes yes yes no no no no no no no no yes yes yes no Chlorpheniramine yes yes no no no yes yes no no no yes yes yes yes no Ibuprofen no no no no no no no yes yes no yes no yes yes no Benzoflumethiazide yes yes no no no no no yes yes yes yes yes yes yes no Tetramisole yes yes yes yes no no no yes yes yes yes yes yes yes yes Sulpiride yes no yes no no no no no no yes no yes no no no Sulconazole yes yes yes yes no no no yes yes yes no no no yes no Octopamine no no no yes no yes no no no no no no no no no Warfarin yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes Propranolol yes yes yes yes yes yes yes yes yes yes yes no no yes yes Ornidazole no no no no no no no no no no no yes yes no yes Acenaphthenol no no no no yes yes yes yes yes yes yes yes yes yes yes trans-stilbene oxide yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes 2-methyl-2-propyl-sulfinamide no no no no no no no no no yes no no no no no Ketoprofen no no no no no no no yes no no no yes yes no yes Miconazole nitrate yes yes yes yes no no no yes no yes yes yes no yes no Naphthol no no no yes yes yes yes yes yes yes no yes no no yes (±)-Naringenin yes yes yes yes yes no yes no no no no no no no no Terfenadine no no no yes yes yes no yes yes yes no yes yes no no Bucetin no no yes no yes yes yes no no no yes no yes no yes Cineole no no no no no no no no no no no no no no no

7 Practical Performance Characteristics Capacity to resolve Percentage of test compound set that interact with CSP R s = RT 2 RT 1 Degree of resolution Capacity to load Overloading decreases peak capacity Measure peak capacity with high and low standard samples Capacity to elute Compounds resolved at lower % co-solvent are easier to recover Some columns are more retentive than others

8 Column Performance Ranking Column Overall rank AD 92.3 Whelk-O OD 78.8 RegisPack 77.8 CLA IC 72.2 IA 70.8 Lux-2 Cel 69.1 CCA-EX-b AS 68.4 OJ 64.7 CC RegisCell 63.7 CCJ 60.8 CCO-EX-b (%RC max +%DR max +%PCL min )/3 Simple, unweighted average of performance criteria Lower ranked columns are more compound specific Greater preference for immobilized CSPs

9 Iterative Screening Initial screen 6 columns, 2-3 co-solvents Stop early if a separation looks promising Second set of columns only if no CSP interaction seen 3 minute 5-55% co-solvent gradient Pick a column, develop isocratic method Co-solvent, %, modifier +/- Transfer to prep column

10 Co-solvent Mediates Selectivity E2 E1 Co-elution E1 Co-solvent: Blue = MeOH Purple = 50:50 MeOH:iPrOH Gold = iproh E2 AV (RegisPack) Reversal of elution order using simple alcohols

11 Optimizing Prep Chromatography Isopycnic plot: CO2/MeOH = 80/20 (v/v) Generated by: REFPROP (NIST) Operating region Critical point: T c = 86 o C, P c = 120 bar, r c =.48 g/ml Limited density variation in operating region gives limited control over solubility and k Courtesy A. Tarafder

12 ABPR Backpressure Optimization ABPR can go unstable, esp. at high pressures, with negative impact on separation Do we need this level of control?

13 Pressure Control ABPR + BPR Cartridge Upchurch 1000 psi cartridge helps stabilize ABPR, but changing pressure is slow Note variation with flow rate and solvent viscosity is minimal

14 Pressure Control Using Restrictor Tubing 30 cm 0.01 ID 20 cm 0.01 ID 10 cm 0.01 ID Selectable restrictor Gilson valvemate + variable restrictor length In practice we use only one restrictor

15 Method Optimization on Prep Columns Vary co-solvent modifier concentration (if used) Adjust retention time using hexane or acetonitrile cosolvent admixtures Elevated temperature (to 40 o C) if retention is excessive Insulated column Set collection times carefully Valley timing to meet ee spec, not for recovery Stack design Larger campaigns are worth more effort

16 E1 E1 E1 E1 E1 Racemic compound separation E2 E2 E2 E2 E2 F1 Valley F2

17 Feedstock: Sample + Impurities UPC2-1 AV15166_ AV15166 Whelk-O (212) 10% 1:1 MeOH/IPA w/ 0.1% IPAm Diode Array 230 Range: Isocratic SFC 1.59 UPLC-1 AV15166 CSH C18 (223) AV15166_06 Sm (Mn, 1x2) 2: Diode Array Range: 3.25 Time Height Area Area% Gradient HPLC AU e-1 AU e-1 4.0e e-1 2.0e e e-1 7.0e-1 Time e-1 5.0e-1 4.0e-1 3.0e-1 72 % chemical purity 2.0e-1 1.0e Time

18 Stacked Injection Design E1 E2 Estimated productivity 0.35 kkd, E1 >95% ee Requires knowing the desired peak

19 Advanced Strategies Resolve a key intermediate Evaluate different feedstocks Derivatize with removable protecting groups Partially enrich one isomer

20 Key Intermediate H 3 C CH 3 O CH 3 H 3 C CH 3 O CH 3 BnOOC N CN i) Bu 3 SnN 3, xylene, reflux ii) HCl, MeOH HOOC N N N N NH iii) NaF, aq. NaOH iv) filter, HCl Proprietary Chiral SFC Method Est. productivity 1.8 kkd Valsartan (Diovan) Chiralpak AD, Hex:IPA:TFA Est. productivity 0.08 kkd

21 Feedstock Evaluation Case Study in purification strategy silica C grams active enantiomer needed By conventional chromatography: 8 FTE-wks to purify racemate SFC: Cleanup on 2-EP at 4.5 g/hour = 1 week to pure rac. Cost savings: client estimate $55K

22 Protection/Deprotection OH OH O H 3 C CH 3 R 2 NH 2 * * Boc 2 O, NaHCO 3 R 2 NH O CH 3 R 1 HCl R 1 Unprotected: Inadequate resolution, esp. on loading

23 Chiral Chromatography of Protected Compounds FMOC: ChromegaChiral CCS, Gradient, ACN: MeOH 3:1 BOC: ChiralPak AD-H, Gradient, IPA

24 BOC-protected (Isocratic Analytical)

25 BOC-protected (Isocratic Preparative) Est. Productivity 2.4 kkd

26 Selective Enantiomer Enrichment E1 production is more efficient than E2 enrichment Standard sample of active Pirkle columns, playing with selectivity, and CSP changes can often reverse elution order Feedstock sources Synthetic: partial success of enantioselective process Chromatographic: overload and polish

27 Co-solvent Changes Selectivity E2 E1 Co-elution Co-solvent: Blue = MeOH Purple = 50:50 MeOH:iPrOH Gold = iproh E1 E2 AV (RegisPack)

28 Selective Enantiomer Enrichment E1 production is more efficient than E2 enrichment Standard sample of active Pirkle columns or phase changes can reverse elution order Enriched feedstock sources Synthetic: partial success of enantioselective process Chromatographic: overload and polish

29 E1 E1 E1 E1 E1 Racemic compound separation E2 E2 E2 E2 E2 F1 Valley F2

30 Enriching E1 Return on Effort %E1 %E2 Processing Rate (kkd) Prod. Rate E1 (kkd) Prod. Rate E2 (kkd) Enriched solutions of warfarin (MeOH, ChiralPak AD-H) Compare feedstock processing rate to get to 95% ee with production rates of each isomer Processing E1-enriched feedstock is 4.3x more efficient than racemic, but enriched E2 is only 2.4x better

31 Multi-step Separations 50 min RP gradient: impractical for preparative HPLC Goal: separation of 10 mg of each target peak for full structure elucidation Tools for separation effort: LCMS/UV(PDA), SFCMS, Prep SFC

32 Fraction 11 QC by HPLC

33 Typical Fraction Tree AV11083 Crude F1 Method 3 F2 Method 1 F3 F4 F9 F10 F11 F5 F6 F7 F8 Method 4 Method 5 Method 7 F12 F13 F32 F33F34 F14 F15 F16 F17 F18 F19 F20 F21 Method 6 Method 7 Method 7 F22 F23 F24 F25 F26 F27 F28 F35 Method 8 F36 F39 F37F38 F40 F41 F29 Method 8 F30 F24-28 F31

34 Standard Fraction Manifold Collect valve Fraction valve F1 UV Detector BPR Heat Exchanger F2 Waste F4 F3 MBPR and manifold 17 bar MBPR waste Waste bottle Fraction bottles & vent lines

35 Waters Prep-100 MBPR Tunable Splitter UV Detector BPR Heat Exchanger GLS MS Detector Reverse Gradient Solvent Pump FC valve Open bed fraction collector

36 Prep-100 GLS provides nearperfect alignment of UV, MS, and actual fraction traces Ebinger, Weller, Kiplinger, Lefebvre; J. Amer. Assoc. Lab. Automation 2010

37 Fraction Collection Using GLS 7 bar Collect valve F1 MBPR F2 UV Detector BPR Heat Exchanger GLS F3 F4 Reduced tubing length post-gls MBPR set very low No need for reversed gradient pump Fraction valve is the fraction vessel Waste Fraction valve

38 Expanded Fraction Collection

39 Initial Cuts from Mother Liquor F1 F4 Seven initial fractions

40 F1 Secondary Fractions Six secondary cuts

41 F4 Fractionation

42 AV UPLC

43 AV15337 HPLC

44 AV SFC

45 AV15337 Prep

46 AV15337 Prep Stack

47 AV UPLC

48 AV15338 HPLC

49 AV SFC

50 AV15338 Prep

51 AV15338 Prep Stack

52 Summary Good enough method development provides a speed advantage that can compensate for sub-optimal separation Work-arounds overcome sub-optimal method issues Drug Discovery working paradigm expects rapid turnaround Problem-solving complex separations with good enough approach, similarly, requires acceptance of sub-optimal approach

53 Thanks Waters Abhijit Tarafder (Waters) Averica John Tipping Brittany Murphy Mickey Rego Keith Galyan Emily-Showell Rouse