TITLE OF THE INVENTION

PROCESS FOR SYNTHESIZING A CETP INHIBITOR

FIELD OF THE INVENTION This invention relates to a process for synthesizing a chemical compound that inhibits cholesterol ester transfer protein (CETP). The product of the process raises HDL- cholesterol in mammals and is expected to have utility in the treatment and/or prevention of atherosclerosis and in delaying the advancement of atherosclerosis.

BACKGROUND OF THE INVENTION

Atherosclerosis and its clinical consequences, coronary heart disease (CHD), stroke and peripheral vascular disease, represent a truly enormous burden to the health care systems of the industrialized world. In the United States alone, approximately 13 million patients have been diagnosed with CHD, and greater than one half million deaths are attributed to CHD each year. Further, this toll is expected to grow over the next quarter century as the average age of the population increases and as an epidemic in obesity and diabetes continues to grow. Inhibition of CETP is a promising new approach to reducing the incidence of atherosclerosis. Statins have been important in reducing the incidence of CHD by reducing LDL- cholesterol (the "bad cholesterol"), but are relatively ineffective at raising HDL-cholesterol ("the good cholesterol"). CETP inhibitors raise HDL-cholesterol, and may provide a potent new tool for reducing CHD and atherosclerosis in the general population. Administration of both a CETP inhibitor and a statin may be especially valuable for treating and preventing atherosclerosis. Pharmaceuticals containing CETP inhibitors are not currently available.

SUMMARY OF THE INVENTION

The present invention provides a process for preparing a family of CETP inhibitors exemplified by the compound having formula I. These novel compounds are potent CETP inhibitors:

The complete process for synthesizing the compound of formula I is summarized in Scheme I, and is subsequently disclosed in detail, step by step.

Scheme 1

R = H ;28 R = Bz ; 27

The process is convergent, and concludes with the coupling reaction of intermediates 29 and 30 to yield the methyl ester of Compound I, followed by the hydrolysis of the methyl ester to yield Compound I as a carboxylic acid.

The compound of Formula I is isolated by the process disclosed herein as a crystalline anhydrous product which is characterized by x-ray powder diffraction, thermogravimetric analysis (TGA), and differential scanning calorimetry (DSC), and by its solid state carbon-13 and fluorine- 19 NMR spectra.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a characteristic X-ray powder diffraction pattern of the crystalline anhydrous form of compound I.

FIG. 2 is a characteristic TGA curve of the crystalline anhydrous form of compound I.

FIG. 3 is a characteristic DSC curve of the crystalline anhydrous form of compound I.

FIG. 4 is a characteristic carbon-13 cross-polarization magic-angle spinning (CPMAS) nuclear magnetic resonance (NMR) spectrum of the crystalline anhydrous form of compound I.

FIG. 5 is a characteristic fluorine- 19 cross-polarization magic-angle spinning (CPMAS) nuclear magnetic resonance (NMR) spectrum of the crystalline anhydrous form of compound I.

DETAILED DESCRIPTION OF THE INVENTION

Further embodiments are described below. Definitions

The terms used throughout this application, and particularly in the examples, are generally well known to chemists who work in the area of process research. Some of these are also defined below:

"EDC" is l-ethyl-3-(3-dimethylaminopropyl)carbodiimide.

"DIP EA" is diisopropylethylamine.

"DMAC" is dimethylacetamide.

"DMSO" is dimethylsulfoxide. "DMF" is dimethylformamide.

"Halogen" includes fluorine, chlorine, bromine and iodine.

"HMPA" is hexamethylphosphoric triamide.

"HOBT" is 1-Hydroxybenzotriazole.

'TP AC" is isopropyl acetate. "Me" represents methyl.

"NaHMDS" is sodium hexamethydisilazide.

"TMEDA" is tetramethylethylenediamine.

"Weinreb amine" is N,O-dimethylhydroxylamine.

EXAMPLE 1

The complete process for making Compound I is described in detail below. Changes in ring substitution to make other compounds that also will be CETP inhibitors can be made by using different substituents on the aromatic rings of the reactants so that Ci-3alkyl, CF3, OCH3, OCF3, F, Cl, and Br are substituted for the Cl, CF3, and CH3 groups that are presently on the aromatic rings of Compound I. These modifications are readily made by one of ordinary skill in the art and are within the scope of the invention.

Furthermore, leaving groups other than the tosylate group of Compound 29 can be used as leaving groups for carrying out the final coupling step to make Compound I. These include but are not limited to Cl, Br, I, benzensulfonate (besylate), methanesulfonate (mesylate), trifluoromethanesulfonate (triflate), and trifluoroacetate. The methyl ester group in Compound 29 can be replaced by other alkyl esters which are also used as protecting groups, such as ethyl, n-propyl, isopropyl, n-butyl, isobutyl, and tert-butyl.

Biaryldichlorobenzyl alcohol 24

A 100 L flask equipped with an overhead stirrer, thermocouple, nitrogen inlet, dropping funnel and a steam pot was charged with 30 L of DMF, 8 L of GMP water, 5000 g of benzyl alcohol and 4458 g of potassium acetate. The mixture was degassed by sparging with nitrogen and then by 3 cycles of vacuum/nitrogen back-fills. PdCl ₂ (DTBPF) (64.8 g) was charged to the reaction mixture and the batch was heated to 40-45 °C. DTBPF is l,l'-bis(di-tert- butylphosphino) ferrocene.

A 20 L flask equipped with an overhead stirrer and nitrogen inlet was charged with 10 L of DMF and 4114 g of dichlorophenyl boronic acid. The solution was degassed by

sparging with nitrogen and then by 3 cycles of vacuum/nitrogen back-fills. A portion of this solution was transferred to the dropping funnel and added slowly to the batch over about 1 h.

The reaction is monitored by HPLC and is complete in ca. 2 h.

Toluene (20 L) and 0.2 M acetic acid (20 L) are added to the reaction mixture. The two-phase mixture is transferred to a 170 L cylinder. Toluene (20 L) and 0.2 M acetic acid (20 L) are added to the reaction flask and this mixture is also transferred to the 170 L cylinder. The layers are separated, and the organic (upper) layer is washed with 10 wt% aqueous sodium chloride (2 x 40 L) and then with GMP water (40 L). The organic phase is then filtered through a bed of silica gel (3 kg), and the silica gel is rinsed with additional toluene (2 x 20 L) until all of the product is recovered. The toluene solution is batch concentrated to a volume of 20 L and then is flushed with 2 x 50 L portions of heptane. The batch volume is adjusted to 60 L and the bath is warmed to completely dissolve any precipitated product.

The batch was seeded with crystals from earlier batches and was allowed to cool to ambient temperature overnight. The batch was cooled to 0 °C and filtered. The wet cake was rinsed with cold heptane (0 °C, 15 L) and dried under nitrogen and vacuum in the funnel, yielding 5667 g of product (90% yield).

Biaryldichlorobenzyl benzoate 25

A 100 L flask equipped with an overhead stirrer, thermocouple, nitrogen inlet, dropping funnel and a steam pot was charged with 51 L of toluene, 4869 mL of triethylamine, 510O g of the biaryl benzyl alcohol from the previous step, and 102 g of 4- dimethylaminopyridine (DMAP). The solution was cooled to 0°C. Benzoyl chloride (2120 mL) was charged to the dropping funnel and added slowly to the batch over ca. 1 h. The addition is exothermic. After the addition was complete, the batch was warmed to ambient temperature to complete the reaction. A thick white slurry forms. The reaction is monitored by HPLC and is complete about 2 hours after warming to ambient temperature.

Aqueous HCl (1.0 M, 20 L) was added to the reaction flask, and the batch was transferred to a 100 L extractor. The lower aqueous layer was separated, and the batch was washed sequentially with 2 x 12 L of l .0 M HCl and then 2 x 12 L of GMP water. The toluene

solution was dried over anhydrous sodium sulfate and then filtered thru a sintered glass funnel. The toluene solution was batch concentrated to a volume of 10 L, and then 51 L of heptane was added while maintaining the temperature at 40-45 ⁰ C. The batch was allowed to cool to ambient temperature overnight. The batch was then cooled to -5 °C and filtered. The wet cake was rinsed with cold heptane (-5 ⁰ C, 16 L) and dried under nitrogen and then vacuum in the funnel, yielding 5740 g of product (85% yield).

Boronic Acid 9

A 3 -Liter round bottom flask equipped with a mechanical stirrer, thermocouple, and addition funnel was charged with 100 g of solid 4-iodo-3-methylbenzoic acid methyl ester and 1.0 L of dry THF. The mixture was cooled to -25 °C, and 218 mL of i-PrMgCl (2M in THF) was added dropwise over 25 min while the internal temperature was maintained at <-15 ⁰ C. The batch was kept at <-10 °C for 1 hr after the addition of the Grignard reagent. Analysis of a hydrolyzed aliquot showed greater than 97% deiodination.

The batch was then cooled to about -20 °C and quenched with trimethyl borate

(77 g). The trimethyl borate reaction is exothermic. The temperature increased to about -4 °C during the addition of the trimethyl borate over 3 min. The resulting solution was aged for 1 h at <0 °C. The batch was then recooled to about -20 °C and further quenched with 1.0 L of IM

H ₃ PO ₄ . This quench was also exothermic, raising the temperature to 3 °C by the end of the quench. The batch was aged at room temperature overnight.

The THF was then removed by distillation at < 45 °C at reduced pressure. The product slurry was allowed to cool to room temperature, and then was filtered. The cake was washed with water (3 X 500 ml) and toluene (2 X 250 mL) and dried under vacuum with nitrogen sweep for 18 h to give the boronic acid 9 (64.3 g, 91%) as an off-white crystalline solid.

Triaryl benzoate 27

A 100 mL Schlenk vessel equipped with a stir bar, nitrogen/vacuum inlet, and septum was charged with boronic acid 9 (6.79 g, 35.3 mmole) and biaryl benzoate 25 (9.45 g). The flask was purged with nitrogen and transferred to a glovebox. A catalyst suspension of bis(acetonitrile)palladium dichloride (107 mg, 0.41 mmole) and 1 ,2-bis(di-t- butylphosphinomethyl) benzene (292 mg, 0.74 mmole) in acetonitrile (35mL) was made as described below and was charged to the Schlenk vessel in a glovebox. The vessel in which the catalyst suspension was made was rinsed with acetonitrile (5 mL); the rinse was transferred into the Schlenk vessel.

The catalyst was made in a nitrogen- filled glovebox by charging bis(acetonitrile)palladium dichloride (107 mg) and l,2-bis(di-t-butylphosphinomethyl)benzene (292 mg) into a vessel equipped with a stir bar. Acetonitrile (35mL) was then charged. The resulting suspension was agitated at ambient temperature for -2 hr prior to use. This suspension is stable for several days, but some decrease in selectivity and conversion is observed with suspensions that have been stored for more than a week. The 1.8:1 ratio of phosphine ligand to Pd is important for achieving high regioselectivity and high conversion.

Aqueous K ₃ PO ₄ (15.Og of 50% w/w K ₃ PO ₄ , 7.5g of K ₃ PO ₄ ) was charged to the resulting thick slurry at ambient temperature. The Schlenk vessel was then sealed, removed from the glovebox, and attached to a nitrogen bubbler. The resulting biphasic mixture was agitated and warmed in an oil bath which was at 55 °C until the amount of unreacted biaryl benzoate remaining was 1.7 LCAP relative to triaryl benzoate product by HPLC analysis (22 hr). Acetonitrile (40 mL) was added at -30 °C, and the bottom aqueous layer separated. The aqueous layer was back-extracted with acetonitrile (3 mL), and this extract was combined with the main organic layer. The reaction mixture was concentrated to -40% of the original volume while maintaining an external temperature and pressure of 40-42 °C /190-200 mbar. The batch was cooled to -30 ⁰ C, and the organic layer was filtered through a sintered glass funnel directly into the crystallization vessel. The reaction vessel was rinsed with MeCN (17 mL), and the rinses were filtered into the reaction vessel. Once the batch cooled, the triaryl benzoate was observed to begin crystallizing out quickly.

The rapidly crystallizing mixture, which was in a 100 mL, 3-neck round-bottom flask equipped with mechanical stirrer, nitrogen inlet/bubbler, and addition funnel, was diluted

with 43 mL of additional CH ₃ CN, giving an assay of ~6 mL CH ₃ CN/g of triaryl benzoate product. Water (25 mL) was added over 60 min at ambient temperature to the thick slurry to give -27 vol% water (relative to MeCN). The suspension was agitated at ambient temperature until the concentration of triaryl benzoate in the supernatant reached about 5.5 g /L by HPLC analysis (overnight age).

The batch was cooled in an ice bath to ~2 °C and agitated for about 2 hours until the concentration of triaryl benzoate in the supernatant reached -1.6 g/L. The suspension was filtered on a sintered funnel and the cake was washed with a total of 46 ml of 75:25 v/v of chilled CH ₃ CN:water as displacement washes. The cake was dried under vacuum and a nitrogen tent at r.t. until a constant weight was obtained. The overall isolated yield of triaryl benzoate for the reaction was -90% (10.8 g, >99.7 LCAP by HPLC).

Triaryl alcohol (28)

To a IOOL flask with overhead stirrer, nitrogen line and temperature probe was added 2kg of the solid triaryl benzoate (27) and dry methanol (20L). A subsurface sparge of the slurry with nitrogen with stirring was carried out for 5 min. Sodium methoxide (30 wt% solution, 210 mL) was added to the slurry, and the reaction mixture was aged at RT until <0.15LCAP starting triaryl benzoate remained (approximately 3-4 h). The reaction mixture became homogeneous about 1 hr before the end of reaction. 5M HCl (250 mL) was added, followed by toluene (10 L) and water (15 L). The phases were then separated, and the organic layer was washed once with water (10L). The batch was then concentrated to remove water and residual methanol.

Triaryl tosylate 29

A 75-L round-bottomed flask equipped with an overhead stirrer, nitrogen inlet, thermocouple and dropping funnel was charged with 1490 g of triaryl alcohol 28 in toluene (22.2 L). TsCl (1232 g) was added to the solution and stirred, followed by addition of a solution of triethylamine (1197 mL), DABCO (19.3 g) and toluene (2.5 L) via addition funnel over a period of 3 minutes. The internal temperature rose to 32 ⁰ C. The reaction was monitored by HPLC, reaching >99.7% conversion within 2 h.

Once the starling triaryl benzoate was consumed, the solution was filtered through a pad of silica gel (1700 g), and the cloudy suspension became a clear solution. The filter cake was washed with 5.0 L of toluene, and the combined washes were added to a 100-L reactor/extractor. The stirred solution was mixed with 6.0 L of 10% w/w NaHSO ₄ (900 g in 9.0 L GMP water) and stirred vigorously. The phases were partitioned, and the aqueous phase was cut (pH = 1). The organic (top) layer was then washed with GMP water (2 x 9.0 L), and the aqueous layers were separated. The pH values of the 2 washes were 1 and 4, respectively.

The organic layer was concentrated to an oil and dissolved in 10-14 volumes of 3- 7% toluene in heptane. The organics were heated until the solution became clear and homogeneous (between 60-80 ⁰ C, depending on toluene concentration), then slowly cooled to rt. Seeding of the solution (with seeds obtained from earlier batches ) was conducted at 50 ⁰ C, and crystal growth was immediately observed. After overnight cooling to rt, the mother liquor was decanted, and the crystals were washed with 10.0 L of 10% toluene/heptane followed by 6.0 L heptane. The bright white solid was dried for 72 h under vacuum in a nitrogen tent. Yields were typically 82-88% tosylate 28 (99.8 LCAP).

Triaryl oxazolidinone methyl ester

To a IOOL flask was added oxazolidinone 30 (1.35kg) and dry DMF (30.8L). The synthesis of oxazolidinone 30 is provided later. After cooling to -15 to -20 ⁰ C, NaHMDS (1.96L of 2M solution) was added, and the mixture was aged 15-30 min. The triaryltosylate 29 (2.2 kg) in DMF was added to the resulting sodium salt of oxazolidinone 30, and the mixture was allowed to warm to 0 to 5 ⁰ C. After the triaryl tosylate was consumed, 2.44 L of 5M HCl was added, followed by 22L of 20% heptane/ethyl acetate. Finally, water (11 L) was added slowly. The layers were separated and then the organic layer was washed with DMF:water twice and then with water twice. The organic layer was assayed for yield and then filtered through a plug of silica gel to remove excess oxazolidinone 30. The solution was then solvent switched to methanol and used in the final step.

Hydrolysis to Free Triaryl Acid Oxazolidinone

THF (12 L) was added to a 75 L flask equipped with an overhead stirrer, thermocouple, nitrogen inlet, dropping funnel and a steam pot that contained ca. 12 L of a methanol solution of the triaryl oxazolidinone methyl ester (2190 g, 3 moles). Hydrogen peroxide (1800 mL of a 35 wt% solution) was added. The addition was exothermic, and the batch temperature increased from 16 °C to 23 ⁰ C.

Lithium hydroxide monohydrate (378 g, 9 moles) was charged to the reaction vessel, and the batch was heated to 60 °C. The reaction was monitored by HPLC and was complete in 15-16 h. The reaction mixture was cooled to 0-10 °C, and aqueous sodium bisulfite (2185 g/ca. 18 L GMP water) was slowly added over 2 hours to quench the hydrogen peroxide.

The addition was very exothermic. The absence of peroxide was confirmed using EM Quant Peroxide test strips.

MTBE (22 L) and GMP water (8 L) were then added, and the batch was transferred to a 100 L extractor. The lower aqueous layer was cut away, and the upper organic layer was washed with 10 wt% brine (2010 g NaCl/ca. 18 L GMP water). The lower layer was cut away and the hazy organic layer was dried over anhydrous sodium sulfate.

The dried organic phase was transferred thru an in-line filter (1 μm) to a 100 L flask attached to a batch concentrator. The MTBE was removed and replaced with cyclohexane. The batch volume was adjusted to 20 L. The batch was heated to 50-65 ⁰ C to completely dissolve any precipitated solids, and then was cooled to ca. 60 °C and seeded with crystals obtained from earlier batches. The batch was allowed to cool to ambient temperature overnight to yield a white solid. The white solid was isolated by filtration, rinsed with 2 x 2 L of cyclohexane, and dried under a nitrogen bag with vacuum.

The solid obtained at this point is a mixture of cyclohexane solvated and anhydrous triaryl oxazolidinone acid. The amount of cyclohexane present is typically 3-5 wt%. This solvate is converted to the anhydrous material by heating in a vacuum oven at a temperature of 110-135 °C, with the exact temperature depending on the design and size of the oven that is used.

Chiral Synthesis of (4S,5R)-5- [3,5-Bis(trifluoromethyl)phenyl] -4-methyl-l ,3-oxazolidin-2- one (30)

The oxazolidinone intermediate 30 is made directly from the chiral starting material CBZ-L-alanine (8) by the 3 -step route shown below. The enantiomer of this compound (4R,5S)-5-[3,5-bis(trifluoromethyl)phenyl]-4-methyl-l,3-oxaz olidin-2-one can be made by an analogous route starting from CBZ-D-alanine.

Step 1 : Conversion of 8 to 9:

CBZ-L-Alanine (6.5 kg, 28.5 mol), HOBT-hydrate (4.8 kg, 34.8 mol), Weinreb amine-HCl salt (3.4 kg, 36.2 mol) and THF (32 L) are charged to a clean flask under nitrogen.

The mixture is cooled to 0-10 ⁰ C and then DIPEA (12.4L) is slowly added at a temperature less than 25°C. EDC-HCl (7Kg, 36.2 mol) is then added slowly with cooling at 15°-25°C. The slurry is aged overnight at 20°-25°C. The mixture is then cooled to 0°-10°C, and 3 N HCl (12L) is added slowly. Then IPAC (32 L) is added and the layers are separated. The organic layer is washed once with HCl (13L) and twice with 8% NaHCO3 (13L) (CAUTION: FOAMING). The organic layer is then concentrated under vacuum to about 15L at 5O ⁰ C. The clear solution is cooled slowly to room temperature, allowing the product to crystallize. Heptane (~70L) is then added slowly. The slurry is filtered, washed with heptane (18L), and dried at room temperature on the filter pot. Product is obtained with >99.9%ee measured by chiral HPLC.

Step 2: Conversion of 9 to 10

10

The Weinreb amide 9 from the previous step (6kg, 22.5 mol) and 3,5- bis(trifluoromethyl)bromobenzene (4.85L, 28.1 mol) are dissolved in anhydrous THF (24L). The solution is purged with nitrogen to remove oxygen. The water content should be <500ppm at this point. Atmospheric distillation can be carried out to azeotropically remove water if necessary. The solution is cooled to -10°C and iso-PrMgCl in THF (56.4 mol) is slowly added (2 hours) to the reaction via addition funnel, maintaining a reaction temperature <-5°C. The solution is allowed to warm to 20°C and aged overnight at 20° C, until the amide is <0.5 LCAP. The reaction is then cooled to -10°C under nitrogen and is quenched slowly over 2 hours into 5N HCl (14L) that is maintained at 0-5°C. MTBE (12L) is added and the biphasic mixture is agitated for 5 min. After warming to 20°-25°C, it is allowed to settle for 30 min, and then the layers are separated. The organic layer is washed with water twice (12L).

The organic layer is vacuum transferred through a 1 -micron in-line PTFE filter into a distillation flask and is then concentrated to ~12L under vacuum (internal temperature <40°C) to a minimum agitated volume. The solution is then azeotropically dried with toluene and taken to a minimum agitated volume again. The solution containing ketone 10 is used directly in the next step.

Step 3: Reduction of Ketone 10 to Chiral Oxazolidinone 30:

The ketone 10 (6 kg) is heated at 50°C with 0.3 eq of Al(O-i-Pr)3 (79Og) in 12L IPA and 18 L of toluene for 15.5 hours. The solution is cooled to ambient temperature, and solid KOH pellets (1.35 kg) are added slowly with vigorous stirring, while keeping the temperature at < 25 °C. After about 2 hours, when HPLC shows > 99.5% cyclization, 33L of IN HCl solution is added to quench the reaction, which is kept at < 25 ⁰ C. If a rag layer of solids forms, it should be filtered off. The rag layer is racemic oxazolidinone, and removal increases the enantiomeric excess. The organic layer is then washed first with 36L of 0.5N HCl, then with 6L IPA combined with 45L water, and finally with 6L IPA combined with 36L water. The organic layer is transferred via an inline filter. The solvent is switched to heptane (target volume is ~42L) at -40 °C until <2 v% toluene is left. Aging at rt for 2 h gives the solid product 11.

EXAMPLE 2

The crystalline anhydrous product (anhydrate) having Formula I that is obtained by the process disclosed above is characterized by its X-ray power diffraction pattern, thermogravimetric analysis (TGA), differential scanning calorimetry (DSC), and its solid state carbon- 13 and fluorine- 19 NMR spectra.

X-ray powder diffraction studies are widely used to characterize molecular structures, crystallinity, and polymorphism. The X-ray powder diffraction pattern of the crystalline anhydrous form (also referred to herein as an anydrate) of Compound I was generated on a Philips Analytical X'Pert PRO X-ray Diffraction System with a PW3040/60 console. A PW3373/00 ceramic Cu LEF X-ray tube emitting K- Alpha radiation (wavelength = 1.54187 A) was used as the source.

Figure 1 shows an X-ray diffraction pattern for the crystalline anhydrous form of Compound I. The crystalline anhydrous product exhibited characteristic diffraction peaks corresponding to 2-theta values of 4.1, 6.5, 8.3 and 13.7 degrees, corresponding to d-spacings of 21.38A", 13.62A", 10.65A", and 6.44A°. The anhydrous product was further characterized by diffraction peaks corresponding to 2-theta values of 14.7, 16.9, 17.4 and 18.3 degrees,

corresponding to d-spacings of 6.03A°, 5.24A°, 5.1OA ⁰ , and 4.85A". The anhydrous product was further characterized by diffraction peaks corresponding to 2-theta values of 19.2, 22.7, and 25.2 degrees, corresponding to d-spacings of 4.63 A°, 3.92A", and 3.54A". A listing of peak positions, d-spacings and relative intensities are provided in Table 1.

Table 1. Peak Positions, d-spacings and Relative Intensities of the crystalline anhydrous form of Compound I.

Position [°2Th.] d-spacing [A] ReI. Int. [%

4.1 21.3766 52.10

6.5 13.62276 60.22

8.3 10.65149 26.65

13.7 6.44224 99.96

14.7 6.0276 34.41

16.9 5.24026 56.23

17.4 5.10184 63.62

18.3 4.85446 100.00

19.2 4.62758 157.66

22.7 3.92022 96.11

25.2 3.53986 29.50

Figure 2 shows a characteristic TGA curve for the crystalline anhydrous form of Compound I. A Perkin Elmer Model TGA 7 or equivalent instrument was used. Experiments were performed under a flow of nitrogen using a heating rate of 10 ⁰ C /min to a maximum temperature of approximately 300 ⁰ C. After automatically taring the balance, 1 to 20 mg of sample was added to the platinum sample pan, the furnace was heated, and the heating program was started. Weight/temperature data were collected automatically by the instrument. Analysis of the results was carried out by selecting the Delta Y function within the instrument software and choosing the temperatures between which the weight loss was to be calculated. Weight losses are reported up to the onset of decomposition/evaporation. A weight loss of 1.1% was observed from ambient temperature to 197.1 ⁰ C that was attributed to removal of residual cyclohexane.

Figure 3 shows a characteristic DSC curve for the crystalline anhydrous form (anhydrate) of Compound I. A TA Instruments DSC 2910 or equivalent instrument was used. Samples weighing between 1 and 6 mg were weighed into an open pan. The pan was then crimped and placed at the sample position in the calorimeter cell. An empty pan was placed at the reference position. The calorimeter cell was closed and a flow of nitrogen was passed through the cell. The heating program was set to heat the sample at a heating rate of 10 ⁰ C /min to a temperature of approximately 300 ⁰ C. The heating program was started. When the run was completed, the data were analyzed using the DSC analysis program contained in the system

software. All endotherms were integrated between baseline temperature points that are above and below the temperature range over which each endotherm is observed. The data reported are the onset temperature, peak temperature and enthalpy. The DSC curve is characterized by a large melting endotherm with onset at 181.8 ⁰ C, peak at 187.7 ⁰ C, and an associated enthalpy of melting of 42.3 J/g. This endotherm is preceded by two endotherms of much lower enthalpy - the first with an onset at 152 ⁰ C, a peak at 159.5 ⁰ C and an associated enthalpy of 0.8 J/g, and the second with an onset at 167.8 ⁰ C, peak at 169.8 ⁰ C and an enthalpy of melting of 0.2 J/g. These endotherms are associated with the presence of residual cyclohexane in the lattice.

The solid-state carbon- 13 NMR spectrum was obtained on a Bruker DSX 500WB

NMR system using a Bruker 4 mm H/X/Y CPMAS probe. The carbon- 13 NMR spectrum utilized proton/carbon- 13 cross-polarization magic-angle spinning with variable-amplitude cross polarization, total sideband suppression, and SPINAL decoupling at 100kHz. The sample was spun at 10.0 kHz, and a total of 1024 scans were collected with a recycle delay of 5 seconds. A line broadening of 10 Hz was applied to the spectra before FT was performed. Chemical shifts are reported on the TMS scale using the carbonyl carbon of glycine (176.03 p.p.m.) as a secondary reference.

Figure 4 shows a characteristic solid-state carbon- 13 CPMAS NMR spectrum of the crystalline anhydrous form (anhydrate) of Compound I. The crystalline anhydrous form exhibited characteristic signals with chemical shift values of 131.8, 137.5, and 76.8 p.p.m. The crystalline anhydrous form is further characterized by signals with chemical shift values of 130.2, 122.7, and 40.7 p.p.m. The crystalline anhydrous form is further characterized by signals with chemical shift values of 128.8 and 13.0 p.p.m. The crystalline anhydrous form (anhydrate) is further characterized by signals with chemical shift values of 127.3 and 134.7 p.p.m. Peak assignments and intensities for 10 peaks in the carbon- 13 spectrum are shown in Table 2 below.

Table 2.

Chemical Shifts and Peak Intensities of C- 13 NMR Spectrum

The solid-state fluorine- 19 NMR spectrum was obtained on a Bruker DSX 500WB NMR system using a Bruker 4 mm H/F/X CPMAS probe. The fluorine- 19 NMR spectrum utilized proton/ fluorine- 19 cross-polarization magic-angle spinning with variable- amplitude cross polarization, and SPINAL decoupling at 62.5kHz. The sample was spun at 15.0 kHz, and a total of 512 scans were collected with a recycle delay of 5 seconds. A line broadening of 10 Hz was applied to the spectrum before FT was performed. Chemical shifts are reported using poly(tetrafluoroethylene) (Teflon®) as an external secondary reference which was assigned a chemical shift of-122 p.p.m.

Figure 5 shows a characteristic solid-state fluorine- 19 CPMAS NMR spectrum for the crystalline anhydrous form (anhydrate) of Compound I. The crystalline anhydrous form exhibits characteristic signals with chemical shift values of- 63.4 and - 65.2 p.p.m., with relative intensities of 100% and 87% respectively.

Utilities

Compound I and other compounds made by the process disclosed herein are inhibitors of CETP and have utility in increasing the amount of HDL-cholesterol and reducing the amount of LDL-cholesterol in a patient, preferably a human patient. Increases of HDL-

cholesterol and reductions of LDL -cholesterol are believe by practitioners in the field of medicine to be advantageous in reducing atherosclerosis and associated diseases. Compound I is a very potent inhibitor of CETP, having an IC50 value of 5OnM using assays described in

WO2007/079186. The crystalline anhydrate of Compound l is a stable crystalline form which can be formulated by conventional methods well known in the art to make a tablet, capsule or other dosage form having a suitable dose for reducing CETP activity in a patient. Such a dose may be in the range of 5mg to 250mg, for example 5mg, lOmg, 25mg, 50mg, lOOmg, 150mg, 200mg, or

250mg. The crystalline anhydrate may also be converted to other forms of the compound, such as pharmaceutically acceptable salts of the carboxylic acid groups, to prepare dosage forms having improved solubility and bioavailability.

The crystalline anhydrate of Compound I and other forms of Compound I can also be made into non-conventional formulations that are designed to improve oral bioavailability compared with conventional formulations. All forms of Compound I can be formulated by dissolving them in oils and/or surfactants or dispersing them as non-crystalline dispersions in water soluble polymers, such as polyvinylpyrrolidone) or copolymers of poly(vinylpyrrolidone). An exemplary formulation of the crystalline anhydrate of Compound I comprises a dose of 5 mg, 10 mg, 50 mg, 100 mg, or 150 mg dissolved in a mixture of surfactants, such as polysorbate 80 and Imwitor 742, to make a solution for use in a gelatin capsule. Such doses may be administered once or twice a day. Such formulations are well known to those of skill in the art of pharmaceutical formulations.

The formulations can be administered to patients having atherosclerosis, such as patients having ACS (atherosclerotic coronary syndrome), to reduce or reverse the progression of atherosclerosis, thereby reducing mortality in a patient population having ACS.