Development of an Efficient Process for a Key Synthetic Intermediate of the SGLT2 Inhibitor LH-1801

Abstract

(2-Bromo-5-chloro-4-((5-ethylthiophen-2-yl)methyl)phenyl)methanol (1) is a key intermediate of the SGLT2 inhibitor LH-1801. This study aimed to explore an efficient process for the synthesis of 1 on a large scale. In this work, the synthetic route started with 3-amino-4-methylbenzoic acid methyl ester (2), followed by (1) bromination, (2) diazotization–Sandmeyer reaction, (3) the oxidation of the benzylic hydrogen to give the corresponding benzoic acid, (4) Friedel–Crafts acylation, and (5) reduction of a carbonyl group (C = O) into a methylene group (–CH₂–) to achieve the target product. We optimized the diazotization-Sandmeyer reaction conditions and reaction parameters for a successful oxidation sequence. In addition, a reduction system for Step 5 was screened. With the optimal reduction system (NaBH₄/BF₃•THF) in hand, the target product (1) could be generated in an efficient and concise one-step reduction approach. Given the above, the synthesis route with process optimization delivered the target product (1) in only five steps with an overall yield of 18.9% and a purity higher than 99%.

 

MissingLinkText

Keywords

SGLT2 inhibitor - LH-1801 - diabetes - sodium borohydride - boron trifluoride

Introduction

Sodium-Glucose Co-Transporter 2 Inhibitor (SGLT2) inhibitors, including Gliflozins, are a very important class of prescription drugs used to treat type II diabetes. However, when exploring an SGLT2 inhibitor, most efforts have been focused on the modification of aglucone,[1] with relatively few modifications to sugar. Liu's research group has designed and synthesized a novel series of 6-deoxy O-spiroketal C-arylglucosides as SGLT2 inhibitors, over 100 compounds. Li's research group has established a drug efficacy screening platform for diabetes candidates, systematically screened over 100 compounds for their in vivo and in vitro activities, and conducted pharmacological evaluation of the candidate new drug LH-1801 in seven animal models.[2] LH-1801 is a novel SGLT2 inhibitor with excellent hypoglycemic effects in vivo for the treatment of type 1 and type 2 diabetes mellitus and is currently undergoing phase III clinical trials. It is a first-class antidiabetic new drug with independent intellectual property rights in China, jointly developed by the Shanghai Institute of Materia Medica, Chinese Academy of Sciences, and Jiangsu Lianhuan Pharmaceutical Co., Ltd.[3]

Jiangsu Faande Pharmaceutical Tech Co., Ltd. used 2-chloro-4-methylbenzoic acid as a raw material to generate 1, a key intermediate of LH-1801 ([Fig. 1]). However, the reported synthetic route suffers from several drawbacks: (1) the high cost of 2-chloro-4-methylbenzoic acid and not commercially available in the market. (2) The overall yield was extremely low, only 2.3%, resulting in high raw material costs and unsuitable for large-scale production.[4]

Fig. 1 Retrosynthesis of SGLT2 inhibitor LH-1801.

Jiangsu Lianhuan Pharmaceutical Co., Ltd. has disclosed a method for obtaining the target compound 1 by using 2-amino terephthalic acid as the raw material, followed by monoesterification, bromination, Sandmeyer reaction, acyl Friedel–Crafts, and reduction. The drawback of this process is that the diazotization reaction is performed in a sulfuric acid medium, where the hydrolysis of the ester is very severe, resulting in a very low yield. The final reduction uses lithium aluminum hydride, which has a high raw material cost and poor process safety, making it unsuitable for future production.[5]

In this paper, we report our efforts on the route scouting toward an efficient synthesis of the key intermediate 1 for the LH-1801 to support clinical development as well as future commercial manufacturing.

 

MissingLinkText

Results and Discussion

Synthesis Route of Compound 1

As shown in [Scheme 1], we describe an efficient five-step sequence toward 1 established during the early development of our program based on a literature screen.[6] [7] [8] [9] The synthesis route started with 3-amino-4-methylbenzoic acid methyl ester (2), and proceeded through a five-step sequence: bromination (step 1),[10] [11] diazotization–Sandmeyer (step 2),[12] [13] oxidation (step 3),[13] [14] Friedel–Crafts acylation (step 4),[15] and reduction (step 5).[16] [17] [18] [19] [20] [21] Based on our knowledge, hydrochloride salt 3 produced in the bromination reaction (step 1) was a new compound that had not been prepared in the literature. In steps 2 and 3, process parameters were optimized based on the literature. In steps 4 and 5, the preparation of 1 from 5 has not been reported in the literature. Herein, we describe our process development toward producing >50 kg scale of 1 with an overall yield of 18.9%. The process optimization is shown below.

Scheme 1 Scale-up synthesis route of 1 proposed in this work for preclinical studies.

 

MissingLinkText

Optimization of the Bromination Sequence to Generate 3

Compound 2 underwent a bromination reaction to obtain methyl 5-amino-2-bromo-4-methylbenzoate (3). Our preliminary study showed that when acetonitrile (10 V), ethyl acetate (5 V), N,N-dimethylformamide (DMF) (10 V), and dichloromethane (DCM, 10 V) were used as reaction solvents, 3 was obtained with significantly lower purity. When DCM (5 V) was used as the reaction solvent, and a solution of NBS in DMF (3 V) was added dropwise, the purity of 3 reached 99.7% ([Supplementary Table S1], available in online version). Then, brominating reagents were screened; our data suggested that NBS was better than DBDMH ([Supplementary Table S2], available in online version); however, when the NBS concentration was reduced, the yield of 3 was significantly increased ([Supplementary Table S3], available in online version). The bromination sequence was successfully performed to produce >450 kg of 3 in six batches with yields >80 and >99.0% high-performance liquid chromatography (HPLC) purity.

 

MissingLinkText

Optimization of the Diazotization–Sandmeyer Reaction Toward 4

Compound 3 underwent a diazotization–Sandmeyer reaction to obtain methyl 2-bromo-5-chloro-4-methylbenzoate (4). During the reaction course, 2-bromo-5-chloro-4-methylbenzoic acid (4a) was generated as a by-product, which may be produced by the hydrolysis of 3 or 4. Thus, the stability of 3 and 4 in the acidic aqueous medium was investigated at low temperature. As shown in [Table 1], compound 3 was relatively stable, while the purity of 4 decreased from 95.90 to 89.94% within 2 hours; therefore, the optimal reaction time was less than 2 hours. Furthermore, we observed an increasing trend in yield along with increasing scale ([Supplementary Table S4], available in online version). This diazotization–Sandmeyer reaction was robustly scaled up to produce approximately 280 kg of compound 4 in 12 batches with an average yield of 72% and >99.0% HPLC purity.

Table 1 Stability studies of 3 and 4 for Sandmeyer reaction[a]

Sampling time	HPLC (area%)	HPLC (area%)
	4	3
0 h	95.90	95.30
1 h	94.91	95.31
2 h	89.94	95.25
7 h	63.38	92.09
24 h	59.07	89.61

Abbreviation: HPLC, high-performance liquid chromatography.

^a Reaction condition: add 1.0 g of samples 3 and 4 to a solution of 1,4-dioxane (5 mL) and 36 wt% HCl (aq) (5 mL), respectively. The mixture was stirred at 0 to 5°C. The sample was monitored by an HPLC method at the set time.

 

MissingLinkText

Optimization of the Oxidation Reaction Toward 5

The phenylmethyl of 4 underwent an oxidation reaction in the presence of KMnO₄/H₃PO₄ to obtain 5-bromo-2-chloro-4-(methoxycarbonyl)benzoic acid (5). The reaction conditions were optimized. First, reaction temperatures (50, 60, 70, and 80°C) were screened. As shown in [Supplementary Table S5] (available in online version), the lower temperature (50, 60°C) inhibited the hydrolysis of methyl ester to 2-bromo-5-chloroterephthalic acid (the corresponding di-acid), and 5 was present with a low yield, significantly reducing the rate of oxidation. When the reaction temperature was 70 and 80°C, the di-acid was remarkably enhanced. Therefore, it would be a challenge to inhibit hydrolysis and improve the conversion rate during the reaction course.

Second, the acids, including benzoic acid, acetic acid, sulfuric acid, boric acid, and phosphoric acid, were screened. As shown in [Table 2], di-acid reached approximately 48% (area%) in the presence of benzoic acid ([Table 2], entry 1); however significantly reduced in the presence of acetic acid, sulfuric acid, and boric acid ([Table 2], entries 2, 3, 4). Fortunately, when phosphoric acid (H₃PO₄) was used, the reaction conversion was accomplished effectively, yielding a considerable amount of 5 while a small amount of di-acid ([Table 2], entry 5).

Table 2 Acid screening for the oxidation reaction toward 5 [a]

Entry	Acid	HPLC (area%)
		Di-acid	5	4
1	Benzoic acid	48.19	0.14	50.27
2	Acetic acid	0.25	0.55	97.15
3	Sulfuric acid	0.84	13.25	82.77
4	Boric acid	0.17	7.40	90.61
5	H3PO4	2.82	80.63	15.30

Abbreviation: HPLC, high-performance liquid chromatography.

^a Reaction condition: a mixture of compound 4 (30.0 g), 18-crown-6 (3.75 g), tert-butanol (180.0 g), and water (555.0 g) was stirred and heated in an oil bath. The internal temperature was 55 to 60°C. Potassium permanganate (99.0 g) was added gradually in batches of 15.0 g. Before adding the next batch, the generated alkali was neutralized with acid. The sample was monitored by an HPLC method at the set time.

During the process of oxidizing methyl to a carboxyl group, since one molecule of potassium hydroxide is produced during the reaction, using phosphoric acid can make the reaction system form a buffering salt, effectively controlling the pH of the reaction within the range of 5 to 8. We found that when the pH is too low, the rate of methyl oxidation to the carboxyl group is extremely slow, while when the pH is too high, the methyl ester is easily hydrolyzed to the carboxyl group.

The progress was monitored until approximately 20% (area%) of the compound 4 remained, and the content of di-acid was approximately 3% (area%) ([Supplementary Table S6], available in online version). Crude product 5 was washed with 9 wt.% EDTA (aq) twice to reduce the content of the di-acid from 3.16 to 0.3% ([Supplementary Table S7], available in online version).

This oxidation reaction was robustly scaled up to produce approximately 160 kg of compound 5 in 11 batches with an average yield of 41 and >98.0% HPLC purity. While based on the consumption of compound 4, the yield was in the range of 56 to 60%.

 

MissingLinkText

Optimization of the Friedel–Crafts Acylation Toward 8

Compound 5 was first transformed into an acyl chloride intermediate (6), then underwent a Friedel–Crafts acylation reaction in the presence of AlCl₃ to obtain methyl 2-bromo-5-chloro-4-(5-ethylthiophene-2-carbonyl)benzoate (8). We compared two methods with different charging sequences of 2-ethylthiophene. The results showed that method B led to a higher reaction conversion ([Table 3]). The Friedel–Crafts acylation was robustly scaled up to produce approximately 150 kg of 8 in 4 batches with an average yield of 70 and >95.0% HPLC purity.

Table 3 Comparison of the adding methods A and B

Entry	Adding method	IPC result (area% of HPLC)
		8	5	7
1	A[a]	63.89	21.00	12.05
2	B[b]	78.93	1.74	15.20

Abbreviations: HPLC, high-performance liquid chromatography; IPC, in-process control.

^a Compound 6 (1.0 equiv.) was added to a solution of dichloromethane (DCM) and 2-ethylthiophene (1.5 equiv.) at 0 to 5°C, then AlCl₃ (3.0 equiv.) was added in batches, the reaction was maintained, and samples were taken for in-process control.

^b Compound 6 (1.0 equiv.) was added dropwise to a solution of DCM and AlCl₃ (3.0 equiv.) at 0 to 5°C, then 2-ethylthiophene (1.5 equiv.) was added dropwise. The reaction was maintained, and samples were taken for in-process control.

 

MissingLinkText

Optimization of the Reduction Sequences Toward 1

Compound 8 has two functional groups that need to be reduced to obtain the final product; the ester group would be reduced to alcohol, and the aryl ketone should be deoxygenated to hydrocarbon. As shown in [Scheme 1], initial attention was paid to the successive reduction of 8 to methyl 2-bromo-5-chloro-4-((5-ethylthiophen-2-yl)methyl)benzoate (9) in the presence of BF₃•OEt₂/Et₃SiH, followed by a successive reduction of 9 to (2-bromo-5-chloro-4-((5-ethylthiophen-2-yl)methyl)phenyl)methanol (1) in the presence of NaBH₄.

The reaction conditions were screened. The data showed that the catalytic effect of BF₃•OEt₂ was better than BF₃•THF in producing a considerable amount of 9 and 1 ([Table 4]); therefore, BF₃•OEt₂ was preliminarily selected as the catalyst for this reaction. The reduction was robustly scaled up in the pilot plant to produce approximately 50 kg of compound 1 in 3 batches with an average yield of 40% and >99.0% HPLC purity.

Table 4 Comparison of the additive reagents for the reduction of compound 8 [a]

Entry	Reductant (equiv.)	Additive (equiv.)	IPC result (area% of HPLC)
			8	1	9
1	Et3SiH (4)	BF3•THF (4)	0.72	1.02	97.34
2	Et3SiH (4)	BF3•OEt2 (4)	0.28	25.19	70.88

Abbreviations: HPLC, high-performance liquid chromatography; IPC, in-process control.

^a Reaction condition: 35 to 40°C; 4 hours.

To our knowledge, the target product was obtained in two steps, resulting in a higher production cost of the target product. We expected that this two-step reaction could be simplified by using a one-pot procedure to achieve a reduction effect and compress the process cycle time. We aimed to develop a process to obtain 1 by a one-pot reaction, which was a challenge.

 

MissingLinkText

Postpilot Process Optimization

Considering that borane does not seem to be able to reduce the ester group, the reasonableness of the reduction of the ester group with NaBH₄, and referring to the similar reactions and their mechanistic explanations for the deoxygenation of the carbonyl group of aryl ketone,[17] [18] [19] [20] [21] we proposed the reduction reaction of the carbonyl group to methylene with borane/boron trifluoride.

We first performed the reaction in the presence of the borane dimethyl sulfide complex. However, the reaction rate was very slow, even at high molar ratios ([Table 5], entries 1–3). We found that borane may not only serve as a reducing agent but also as a weak Lewis acid itself. If the borane is replaced with a stronger Lewis acid, such as BF₃, the conversion rate may be improved. This encourages us to find a way to improve the reaction.

Table 5 Scouting reduction systems for the reduction reaction of compound 8

Entry	Reaction conditions				IPC result (area% of HPLC)
	Reductant (equiv.)	Additive (equiv.)	Temperature (°C)	Time (h)	/	/	/	1
1	Borane* (5.2)	/	60–65	45	/	0.64	/	83.30
2	Borane (6.2)	/	60–65	46	/	0.39	/	87.42
3	Borane (3.1)	/	60–65	49	/	0.86	/	83.71
4	NaBH4 (6.0)	CF3COOH (6.0)	50–55	4	79.63	9.76	/	3.04
5	Borane (3.5)	BF3•OEt2 (3.1)	50–55	27	/	/	0.36	91.73
6	Borane (3.5)	BF3•THF (3.1)	50–55	23	/	/	0.31	95.73
7	NaBH4 (3.0)	BF3•THF (6.5)	50–55	12	/	0.32	0.02	96.20

Abbreviations: HPLC, high-performance liquid chromatography; IPC, in-process control; Temp., temperature.

Note: Borane * is the borane dimethyl sulfide complex.

When the additive (BF₃•THF or BF₃•OEt₂) was added to the reaction, the reaction rate was significantly accelerated ([Table 5], entries 5, 6). Then, NaBH₄/BF₃•THF was selected in terms of process safety and cost. NaBH₄ could reduce the ester group and ketone both to alcohol first, then add BF₃•THF to the reaction mixture via borane formed in situ to finish the reduction. This protocol functioned and gave a clean reaction ([Table 5], entry 7). If CF₃COOH were used as a stronger Lewis acid, the product 1 would be less generated ([Table 5], entry 4).

Compared with our initial route, the one-pot reduction approach of the key intermediate 1 from 8 avoids the use of hazardous ethyl ether solvent, rendering the process safer, with a reduction of PMI from 165.7 to 77.9, an increase in the yield from 40 to >60%, and a 45% reduction in cost ([Table 6]). The one-step reduction approach of the key intermediate 1 from 8 was efficient and concise, which was our second-generation process.

Table 6 Comparing the first- and second-generation process of 1

Number	Criteria	First-generation process	Second-generation process
1	Step count	2	1
2	Reaction reagents	Stage 1: Et3SiH/BF3•OEt2 Stage 2: NaBH4	NaBH4 and BF3•THF
3	Yield	About 40%	>60%
4	Hazardous reagents	Ethyl ether	n/a
5	Cost %	n/a	−45%
6	PMI	165.7	77.9

Abbreviations: PMI, process mass intensity; n/a, none.

 

MissingLinkText

 

MissingLinkText

Conclusion

In summary, we have reported an efficient synthesis route for the production of 1, which is an important LH-1801 intermediate. The highlights of this synthesis include: (1) preparation of compound 3 was an efficient process for separating and removing impurities; (2) timely in-process control and separation of products were very critical for the diazotization–Sandmeyer reaction; (3) the one-step reduction approach was successfully deployed with the yield increasing from 40 to >60%. However, the oxidation with potassium permanganate gave a low yield and high pressure for the environment, and the workup process in many steps was tedious. If there are sufficient business or technical driving factors, the team will continue to pursue further improvements, including alternative synthetic routes to get more streamlined processes, efficient isolation protocols, lower PMI, and shorter cycle times.

 

MissingLinkText

Experimental Section

General

Solvents and reagents were bought from commercial suppliers and used without further purification. All reactions were performed in glass-lined reaction vessels equipped with mechanical stirring, a thermometer, and a nitrogen/vacuum inlet in the pilot-plant or laboratory, unless specified otherwise.

¹H (500 MHz) and ¹³C (126 MHz) NMR spectra were recorded on a Bruker AVANCE 500 spectrometer. Chemical shift (δ) is expressed in units of parts per million (ppm) using tetramethylsilane (TMS) as an internal standard and referenced to the solvent peak (DMSO-d ₆: 2.50 ppm for ¹H NMR and 39.53 ppm for ¹³C NMR or chloroform-d: 7.26 ppm for ¹H NMR and 77.1 ppm for ¹³C NMR). Multiplicities are abbreviated as singlet (s), doublet (d), triplet (t), quartet (q), and multiplet (m). Coupling constants (J) are reported in Hertz (Hz). High-resolution mass spectra (HRMS) were taken on AB 5600 (AB SCIEX, United States).

Reaction progress and the purity of the compound were determined by HPLC on Agilent 1260II (Agilent Technologies, California, United States). HPLC method was performed on a YMC-Pack ODS-AQ column (25 cm × 4.6 mm × 5.0 μm). The gradient program was as follows: 0 minute hold with 35/65% (v/v) 0.1% phosphoric acid in water/methanol, then gradient elution from 35/65% (v/v) 0.1% phosphoric acid in water/methanol to 90/10% (v/v) 0.1% phosphoric acid in water/methanol over 25 minutes, hold for 10 minutes, then from 90/10% (v/v) 0.1% phosphoric acid in water/methanol to 35/65% (v/v) 0.1% phosphoric acid in water/methanol with 9 minutes equilibration time. The flow rate was 1.0 mL/min. The column temperature was 30°C. The detection wavelength was 230 nm.

 

MissingLinkText

Preparation of Methyl 5-amino-2-bromo-4-methylbenzoate Hydrochloride (3) and Analytical Data

A thermometer, a nitrogen/vacuum inlet, and DCM (200 kg) were charged in a 1,000 L reactor with mechanical stirring. Compound 2 (1.00 equiv., 0.33 kmol, 55 kg) was added at 20 to 25°C. The mixture was cooled to 0 to 5°C. A solution of NBS (1.07 equiv., 0.35 kmol, 319 kg, 19.7%w/w in DMF) was added over ca. 4.5 hours at 0 to 5°C, and stirring was continued for 0.5 hours at 0 to 5°C. When the content of 2 was ≤ 1.0% monitored by an HPLC method, the mixture was diluted with water (500 kg). The aqueous phase was extracted with DCM (135 kg). The organic phase was combined, washed with water (100 kg), and treated with 36 wt.% HCl (aq) (1.50 equiv., 0.49 kmol, 50 kg) to attain pH = 1 to 2 at 0 to 5°C. The mixture was filtered. The solids were collected, which were washed with DCM (75 kg) and dried at 55 to 60°C under a vacuum (P ≤ −0.08 MPa) to give 3 as a white solid (81.5 kg, 99.3% HPLC purity, 87.2% isolated yield in one of six manufacturing batches). mp: 193.0-196.5°C. ¹H NMR (500 MHz, DMSO-d ₆) δ 7.73 (s, 1H), 7.65 (s, 1H), 3.85 (s, 3H), 2.34 (s, 3H). ¹³C NMR (126 MHz, DMSO-d ₆) δ 165.70, 136.54, 136.10, 135.31, 130.16, 123.80, 53.07, 34.53, 17.44. HRMS (m/z) calcd. for C₉H₁₁BrNO₂ ⁺ [M + H]⁺ 243.9895; found: 243.9955.

 

MissingLinkText

Preparation of 2-Bromo-5-chloro-4-methylbenzoic acid Methyl Ester (4) and Analytical Data

A thermometer, a nitrogen/vacuum inlet, 36 wt.% HCl (aq) (201 kg) and 1,4-dioxane (176 kg) were charged in a 500 L reactor at 20 to 25°C with mechanical stirring. The mixture was cooled to 0 to 5°C. Compound 3 (1.00 equiv., 0.12 kmol, 34.7 kg) was added at 0 to 5°C. A solution of sodium nitrite ( 1.08 equiv., 0.13 kmol, 9.1 kg.) in water (16 kg) was added over ca. 2 hours, and stirring was continued for 0.5 hours at 0 to 5°C. When the content of 3 was ≤ 1.0% monitored by an HPLC method, the reaction was completed, and the diazonium salt of 3 was obtained.

A thermometer, a nitrogen/vacuum inlet, and 36 wt.% HCl (aq) (91 kg) was added to a 1,000 L reactor at 20 to 25°C with mechanical stirring. The mixture was stirred. Cuprous chloride (1.18 equiv., 0.15 kmol, 14.6 kg) and copper powder (1.3 kg) were added. The mixture was cooled to −5 to 0°C. Diazonium salt of 3 was added over ca. 5 hours, and stirring was continued for 1 hour at 0 to 5°C. When the content of diazonium salt of 3 was less than 1.0% monitored by an HPLC method, the mixture was diluted with water (210 kg) at 0 to 5°C for 0.5 hours. The suspension was centrifuged to give a wet solid, which was charged in the reactor in combination with DCM (240 kg) for complete dissolution. Water (50 kg) was added, and stirring was continued for 0.5 hours at 20 to 30°C. The mixture was settled for 0.5 hours. The phases were separated. The organic phase was washed with 5 wt.% NaHCO₃ (aq) (45 kg). The organic phase was collected and distilled at 35 to 40°C under a vacuum (P ≤ −0.08 MPa) until no solvent flowed out. tert-Butanol (43 kg) and water (40 kg) were added. The mixture was heated (∼50-55°C) for 0.5 hours for complete dissolution. The solution was then cooled to 0 to 5°C slowly and stirred for at least 2 hours. After filtration, the solids were dried at 35 to 40°C under a vacuum (P ≤ −0.08 MPa) to give 4 as a white solid (24.6 kg, 99.1% HPLC purity, 75.4% isolated yield in 1 of 12 manufacturing batches).

 

MissingLinkText

Preparation of 5-Bromo-2-chloro-4-(methoxycarbonyl)benzoic Acid (5) and Analytical Data

A thermometer, a nitrogen/vacuum inlet, tert-butanol (194 kg), and water (600 kg) were charged in a 1,000 L reactor at 20 to 25°C with mechanical stirring. 18-Crown-6 (4.04 kg) and 4 (1.0 equiv., 122.57 mol, 32.3 kg) were added. The mixture was heated to 50 to 60°C and stirred. Potassium permanganate (5.5 equiv., 677.10 mol, 16.0 kg × 6 + 11.0 kg, a total of 107.0 kg) was added in batches once per hour. Phosphoric acid (3.8 kg) was added to neutralize the generated potassium hydroxide. The next batch of potassium permanganate (16 kg) was added. The last addition was 11 kg of potassium permanganate and 3.4 kg of phosphoric acid. Potassium permanganate was added over ca. 7 hours, and stirring was continued for 6 hours at 50 to 60°C. When the compound 4 was between 15 and 30% monitored by an HPLC method, the reaction was cooled to 20 to 25°C, and quenched with sodium bisulfite (0.9 equiv., 110.50 mol, 11.5 kg). DCM (172 kg) was added at 20 to 30°C. The mixture was filtered. The filtrate was settled for 0.5 hours and extracted once with DCM (90 kg). The organic phase was collected and distilled under a vacuum (P ≤ −0.08 MPa) below 40°C to give 4 (6.5 kg). The aqueous phase was adjusted to pH = 1 to 2 with 32 wt.% HCl (aq) (∼29 kg), with white solids being precipitated. The mixture was cooled to 5 to 10°C, stirred for at least 1 hour, and filtered. The solids were collected, washed twice with 9 wt.% EDTA (aq) (55 kg × 2) until the di-acid was less than 1.0%, and dried at 40 to 50°C under a vacuum (P ≤ −0.08 MPa) to give the crude product of 5 as a white solid.

A thermometer, a nitrogen/vacuum inlet, the crude product of 5 and isopropanol (38 kg) were charged in a 500 L reactor with mechanical stirring. The mixture was heated (∼60–65°C) for 1 hour for dissolution. The solution was filtered at 60 to 65°C. The filtrate was collected, and then n-heptane (34 kg) was added. The mixture was cooled to 0 to 5°C to keep crystallization for 1 to 2 hours. The mixture was filtered, the solids were dried at 45 to 50°C under a vacuum (P ≤ −0.08 MPa) to give 5 as a white solid (15.0 kg, 99.2% HPLC purity, 41.7% isolated yield in one of eleven manufacturing batches). mp: 229.5-231.5°C. ¹H NMR (500 MHz, DMSO-d ₆) δ 7.81 (d, J = 1.4 Hz, 2H), 3.87 (s, 3H). ¹³C NMR (126 MHz, DMSO-d ₆) δ 166.22, 165.16, 141.85, 134.63, 133.34, 132.42, 130.02, 118.75, 53.33. HRMS (m/z) calcd. for C₉H₇BrClO₄ ⁺ [M + H]⁺ 292.9211, found: 292.9034.

 

MissingLinkText

Preparation of Methyl 2-bromo-5-chloro-4-(5-ethylthiophene-2-carbonyl)benzoate (8) and Analytical Data

A thermometer, a nitrogen/vacuum inlet, 5 (1.00 equiv., 155.0 mol, 45.5 kg), DMF (0.25 kg) and DCM (340 kg) were charged in a 500 L reactor at 20 to 25°C with mechanical stirring. Oxalyl chloride (1.22 equiv., 189.1 mol, 24.0 kg) was added over ca. 1 hour at 20 to 25°C. The reaction mixture was warmed to 30 to 35°C, and stirring was continued for 1 hour. When the content of compound 5 was ≤0.5% monitored by an HPLC method, the mixture was distilled in vacuum (P ≤ −0.09 MPa) below 40°C to give a residue, which was dissolved in DCM (130 kg). Repeat the distillation and dissolution with DCM (130 kg) once again. The resulting mixture was distilled to give a residue, which was dispersed in DCM (340 kg) to obtain a solution of 6 in DCM.

A thermometer, a nitrogen/vacuum inlet, and DCM (340 kg) were charged in a 1,000 L reactor at 20 to 25°C with mechanical stirring, followed by the addition of aluminum trichloride (2.50  equiv., 386.2 mol, 51.5 kg). The mixture was cooled to 0 to 5°C. The solution of 6 in DCM was added at 0 to 5°C over ca. 2 hours, and stirring was continued for 1 hour. 2-Ethylthiophene 7 (1.72 equiv., 267.4 mol, 30.0 kg) was added at 0 to 5°C over ca. 2 hours. The reaction mixture gradually turns brown, and stirring was continued for 8 hours at 20 to 25°C. The reaction was monitored by an HPLC method. When the content of compound 6 was below 1.0%, the reaction was completed.

A thermometer, a nitrogen/vacuum inlet, 32 wt.% HCl (aq) (66 kg) and water (297 kg) were charged in another 1,000 L reactor at 20 to 25°C, with mechanical stirring. The mixture was cooled to 0 to 5°C, and the reaction mixture of the previous step was added over ca. 2 hours, and stirring was continued for 15 minutes at 0 to 10°C. The aqueous phase was separated and extracted with DCM (164 kg). The organic phase was washed with water (124 kg × 2) twice and distilled under a vacuum (P ≤ −0.08 MPa) at 35 to 40°C to give a residue, which was dissolved in isopropanol (53 kg) and n-heptane (169 kg). The mixture was heated (∼60°C) for complete dissolution. The solution was cooled to 0 to 5°C, stirred for at least 1 hour, and filtered. The solids were the crude product 8.

The crude product 8, isopropanol (46 kg), and n-heptane (162 kg) were added to another 500 L reactor with mechanical stirring and a nitrogen/vacuum inlet. The mixture was heated (∼60°C) for 1 hour until dissolution. The solution was then cooled slowly to 0 to 5°C, and kept the crystallization for 1 hour. After filtration, the solids were dried at 45 to 50°C under a vacuum (P ≤ −0.08 MPa) to give 8 as a yellow solid (43.5 kg, 96.6% HPLC purity, 72.4% isolated yield in one of four manufacturing batches). mp: 90.5-91.2°C. ¹H NMR (500 MHz, chloroform-d) δ 7.90 (s, 1H), 7.71 (s, 1H), 7.25 (d, J = 3.9 Hz, 1H), 6.87 (d, J = 3.9 Hz, 1H), 3.98 (s, 3H), 2.93 (q, J = 7.5 Hz, 2H), 1.36 (t, J = 7.5 Hz, 3H). ¹³C NMR (126 MHz, chloroform-d) δ 183.99, 164.86, 160.97, 142.07, 139.78, 137.10, 134.12, 132.75, 130.34, 125.71, 119.71, 52.98, 24.31, 15.43. HRMS (m/z) calcd for C₁₅H₁₃BrClO₃S⁺ [M + H]⁺ 386.9452, found: 386.9444.

 

MissingLinkText

Preparation of (2-Bromo-5-chloro-4-((5-ethylthiophen-2-yl)methyl)phenyl)methanol (1) and Analytical Data

First-Generation Process

A thermometer, a nitrogen/vacuum inlet purged with nitrogen, 8 (1.00 equiv., 93.89 mol, 36.4 kg), acetonitrile (160 kg), and DCM (203 kg) were charged into a 500 L reactor at 20 to 25°C with mechanical stirring. The mixture was cooled to 0 to 5°C. Triethylsilane (5.00 equiv., 469.56 mol, 54.6 kg) was added at 0 to 5°C over ca. 1.5 hours, and stirring was continued for 0.5 hours. Boron trifluoride etherate (5.00 equiv., 469.92 mol, 66.7 kg) was added at 0 to 5°C over ca. 2.5 hours, and stirring was continued for 1 hour. The reaction mixture was warmed to 35 to 40°C slowly. Stirring was continued for another 4 hours. When the content of 8 was ≤ 0.5% monitored by an HPLC method, the reaction was cooled to 0 to 5°C, water (9.2 kg) was added over ca. 1.5 hours, and stirring was continued for 1 hour below 20°C. The mixture was distilled under a vacuum (P ≤ −0.08 MPa) below 40°C to give a residue, which was dissolved in DCM (250 kg) and water (187 kg), and stirring was continued for 1 hour at 20 to 30°C. The mixture was settled for 1 hour. The aqueous phase was separated and extracted with DCM (100 kg). The organic phase was distilled in vacuum (P ≤ −0.08 MPa) at 30 to 40°C to give a residue, which was dissolved in THF (322 kg) to obtain a solution of 9 in THF.

A thermometer, a nitrogen/vacuum inlet purged with nitrogen, and a solution of 9 in THF were charged in a 1,000 L reactor at 20 to 25°C with mechanical stirring. Sodium borohydride (5.04 equiv., 473.17 mol, 17.9 kg) was added in batches over ca. 1 hour. The reaction mixture was warmed to 60 to 70°C slowly, and stirring was continued for 6 hours. When the content of 9 was below 1.0% monitored by an HPLC method, the reaction mixture was cooled to 0 to 5°C. Methanol (25 kg) was added over ca. 5 hours, and stirring was continued for 1 hour at 0 to 10°C. (Caution: gases are generated during the quenching process.) The mixture was distilled in vacuum (P ≤ −0.08 MPa) at 40 to 50°C to give a residue, which was dissolved in water (145 kg). 32 wt.% HCl (aq) (ca. 30 kg) was added to adjust pH to 7 to 8, followed by the addition of DCM (100 kg). The stirring was continued for 1 hour at 20 to 25°C. The mixture was settled for 1 hour. The organic phase was separated, water (150 kg) was added, and stirring was continued for 0.5 hours at 20 to 30°C. The mixture was settled for 1 hour, the organic phase was separated and distilled in vacuum (P ≤ −0.08 MPa) at ≤40°C to give a residue, which was dissolved in n-heptane (200 kg). Active carbon (1 kg) and Celite (1 kg) were added. The mixture was heated (75–80°C) for 1 hour and filtered at 75 to 80°C. The filtrate was collected, cooled down to 0 to 10°C, and stirred for at least 1 hour. The mixture was filtered to give product 1 (∼34 kg) as a wet crude.

A thermometer, a nitrogen/vacuum inlet, the crude product 1 (ca. 34 kg), and ethanol (90 kg) were charged into another 200 L reactor with mechanical stirring. The mixture was heated (∼60°C) for 1 hour for dissolution, then cooled slowly to −5 to 0°C, and allowed to crystallize for 2 hours. The solids were filtered and dried under a vacuum (P ≤ −0.08 MPa) at 45 to 50°C to give 1 as a white solid (13.0 kg, 99.4% HPLC purity, 40.3% isolated yield based on 8 in one of three manufacturing batches). mp: 85.0-86.5°C. ¹H NMR (500 MHz, chloroform-d) δ 7.50 (s, 1H), 7.40 (s, 1H), 6.63 to 6.58 (m, 2H), 4.67 (s, 2H), 4.14 (s, 2H), 2.78 (q, J = 7.5 Hz, 2H), 2.18 (s, 1H), 1.27 (t, J = 7.5 Hz, 3H). ¹³C NMR (126 MHz, chloroform-d) δ 146.68, 139.64, 139.19, 138.25, 134.11, 133.27, 129.35, 125.44, 123.00, 120.11, 64.26, 33.26, 23.48, 15.83. HRMS (m/z) calcd. for C₁₄H₁₅BrClOS⁺ [M + H]⁺ 344.9710, found: 344.9707.

 

MissingLinkText

Second-Generation Process

THF (50 mL) and 8 (1.0 equiv., 0.026 mol, 10.00 g) were added to a 500 mL reactor under nitrogen at 20 to 25°C. Sodium borohydride (2.6 equiv., 0.067 mol, 0.51 g×5) was added in batches once every 0.5 hours over ca. 2.5 hours, and stirring was continued for 1 hour at 20 to 25°C. Boron trifluoride tetrahydrofuran (6.5 equiv., 0.169 mol, 23.67 g) was added over ca. 1 hour below 20°C (Caution! control the drop rate to prevent material rushing.), and stirring was continued for 1 hour at 10 to 20°C. The reaction mixture was warmed to 50 to 55°C, and stirring was continued for 10 hours. When the content of 8 ≤ 1.0% monitored by an HPLC method, the reaction was cooled to 10 to 20°C, methanol (30 mL) was added over ca. 1.5 hours, and stirring was continued for 6 hours below 20°C. (Caution! Gases are generated during the quenching process.)

The mixture was distilled under a vacuum (P ≤ −0.08 MPa) at 35 to 40°C to give a residue, which was dissolved in DCM (50 mL) and water (50 mL). 5 wt.% NaHCO₃ (aq) (50 mL) was added to adjust pH = 7 to 8. The aqueous phase was separated and extracted with DCM (50 mL). The organic phase was washed sequentially with 5 wt.% NaHCO₃ (aq) (50 mL) and 5 wt.% NaCl (aq) (50 mL), and then distilled in vacuum (P ≤ −0.08 MPa) at 35 to 40°C until no solvent flowed out. n-Heptane (50 mL) and isopropanol (5 mL) were added. The mixture was heated (∼65°C) for complete dissolution, then cooled to 5 to 10°C and stirred for at least 1 hour. The solids were filtered and dried in a vacuum oven (P ≤ −0.08 MPa) at 45 to 50°C to give compound 1 as a white solid (6.11 g, 68.5% yield, 99.4% HPLC purity). mp: 85.6-87.0°C. ¹H NMR (400 MHz, chloroform-d) δ 7.53 (s, 1H), 7.43 (s, 1H), 6.82–6.48 (m, 2H), 4.70 (d, J = 4.6 Hz, 2H), 4.17 (s, 2H), 2.80 (q, J = 7.5 Hz, 2H), 2.13 (t, J = 4.6 Hz, 1H), 1.30 (t, J = 7.5 Hz, 3H). ¹³C NMR (101 MHz, chloroform-d) δ 146.7, 139.6, 139.2, 138.2, 134.1, 133.3, 129.4, 125.4, 123.0, 120.1, 64.3, 33.2, 23.5, 15.8.

 

MissingLinkText

Discuss the Process Mass Intensity of the Process

The American Chemical Society Green Chemistry Institute's Pharmaceutical Roundtable has chosen process mass intensity (PMI) as the key, high-level metric for evaluating and benchmarking progress toward more sustainable manufacturing.[22] For the first-generation process: the PMI of step 5 was 165.1. For the second-generation process, the PMI of step 5 was 77.97, compared with the first-generation reduction process, in which the PMI was reduced by half. The PMI of step 1 was 17.59. The PMI of step 2 was 47.65. The PMI of step 3 was 96.28. The PMI of step 4 was 67.42. For the entire second-generation process, the PMI of the five steps from starting material 2 was 579.61, and 31.2% of which was recyclable solvent (DCM and 1,4-dioxane).

 

MissingLinkText

 

MissingLinkText

 

MissingLinkText

Supplementary Materials

Screening of reaction conditions for bromination (, diazotization–Sandmeyer, and oxidation reaction ([Supplementary Tables S1]–[S7], available in online version); HRMS, ¹H NMR, ¹³C NMR, and HPLC spectra for compounds 3, 5, 8, and 1 ([Supplementary Figs. S1–S12], available in online version); as well as HPLC method for compounds 3, 4, 5, 8, and 1 ([Supplementary Figs. S13–S17], available in online version), can be found in the [Supplementary Materials].

 

MissingLinkText

 

MissingLinkText

Conflict of Interest

None declared.

Acknowledgments

We thank the Chemical Research and Development group at Jiangsu Lianhuan Pharmaceutical Co., Ltd. for their full support, achieving the unprecedented speed of development of a key intermediate 1.

• References

• 1 Zhang Y, Ma X, Shan XH, Zhang XW, Li JQ, Liu Y. Novel and practical industrial process scale-up of 5-bromo-2-chloro-4-(methoxycarbonyl)benzoic acid, a key intermediate in the manufacturing of therapeutic SGLT2 inhibitors. Pharmaceut Fronts 2022; 4 (04) e244-e249
• 2 Wang Y, Lou Y, Wang J. et al. Design, synthesis and biological evaluation of 6-deoxy O-spiroketal C-arylglucosides as novel renal sodium-dependent glucose cotransporter 2 (SGLT2) inhibitors for the treatment of type 2 diabetes. Eur J Med Chem 2019; 180: 398-416
• 3 Liu H, Li J, Wang J. et al. C, O-spiro aryl glycoside compound and preparation and application thereof [in Chinese]. CN Patent 106317068 A. January 11, 2017
• 4 Chen G, Liu T, Xi J. LH-1801 intermediate as well as preparation method and application thereof [in Chinese]. CN Patent 113429379 A. September 24, 2021
• 5 Xia C, Niu B, Zhong L, Wang D, Yu B, Jia Z. SGLT2 inhibitor key intermediate as well as preparation method and application thereof [in Chinese]. CN Patent 115819398 A. March 21, 2023
• 6 Lv B, Xu B, Feng Y. et al. Exploration of O-spiroketal C-arylglucosides as novel and selective renal sodium-dependent glucose co-transporter 2 (SGLT2) inhibitors. Bioorg Med Chem Lett 2009; 19 (24) 6877-6881
• 7 Lv B, Feng Y, Dong J. et al. Conformationally constrained spiro C-arylglucosides as potent and selective renal sodium-dependent glucose co-transporter 2 (SGLT2) inhibitors. ChemMedChem 2010; 5 (06) 827-831
• 8 Liu YH, Fu TM, Ou CY. et al. Improved preparation of (1S,З'R,4'S,5′S,6'R)-5-chloro-6-[(4-ethylphenyl)methyl]-3′,4',5′,6′-tetrahydro-6′-(hydroxymethyl)-spiro[isobenzofuran-1(3H), 2′-[2H]pyran]-3′,4′,5′-triol. Chin Chem Lett 2013; 24 (02) 131-133
• 9 Liu YH, Fu TM, Chen ZD, Ou CY. A new and improved process for C-aryl glucoside SGLT2 inhibitors. Monatsh Chem 2015; 146 (10) 1715-1721
• 10 Liu J, Cai J, Tang J, Wang, Wang L, Wang J. Substituted triazole compound, pharmaceutical composition containing substituted triazole compound, preparation method and application of substituted triazole compound [in Chinese]. CN Patent 110655503 A. January 7, 2020
• 11 Yang Q, Sheng M, Huang YL. Potential safety hazards associated with using N,N-dimethylformamide in chemical reactions. Org Process Res Dev 2020; 24 (09) 1586-1601
• 12 Chen Y, Feng Y, Xu B. et al. Benzylic glycoside derivatives and methods of use [in Chinese]. CN Patent 101652377 A. February 17, 2010
• 13 Prous JR, Serradell N, Munoz R. et al. Pyrano[3,2-c][2]benzopyran-6(2H)-one derivatives and uses thereof.. WO Patent 2013045495 A1. April 4, 2013
• 14 Akama T, Balko TW, Defauw JM. et al. Isoxazoline derivatives used in the control of ectoparasites. WO Patent 2013078071 A1. May 30, 2013
• 15 Resconi LMC, Fait A, Sablong R, Izmer VV, Kononovich DS, Voskoboynikov AZ. Catalyst System. CN Patent 113906059 A. January 7, 2022
• 16 Frick W, Glombik H, Theis S, Elvert R. Novel aromatic fluoroglycoside derivatives, pharmaceuticals comprising said compounds and the use thereof. US Patent 2011059910 A1. March 10, 2011
• 17 Fujii H, Yoshimura T, Kamada H. Regioselective pyrrole synthesis from asymmetric β-diketone and conversion to sterically hindered porphyrin. Tetrahedron Lett 1997; 38 (08) 1427-1430
• 18 Fry JL, Orfanopoulos M, Adlington MG, Dittman Jr WP, Silverman SB. Reduction of aldehydes and ketones to alcohols and hydrocarbons through use of the organosilane-boron trifluoride system. J Org Chem 1978; 43 (02) 374-375
• 19 Smonou I. One step reduction of diaryl ketones to hydrocarbons by etherated boron trifluoride-triethylsilane system. Synth Commun 1994; 24 (14) 1999-2002
• 20 Orfanopoulos M, Smonou I. Selective reduction of diaryl or aryl alkyl alcohols in the presence of primary hydroxyl or ester groups by etherated boron trifluoride-triethylsilane system. Synth Commun 1988; 18 (08) 833-839
• 21 Qu J, Gao D, Zhang P. Preparation method of LH-1801 key intermediate. CN Patent 119143719 A. December 17, 2024
• 22 Jimenez-Gonzalez C, Ponder CS, Broxterman QB, Manley JB. Using the right green yardstick: Why process mass intensity is used in the pharmaceutical industry to drive more sustainable processes. Org Process Res Dev 2011; 15 (04) 912-917

Publication History

Received: 26 March 2025

Accepted: 29 July 2025

Article published online:
20 August 2025