Abstract
The process selection and subsequent development of a reliable, scalable synthesis of the anticancer prodrug tirapazamine (SR259075) is described in this paper. Reaction of benzofuroxan with cyanamide in acetonitrile in the presence of 1,8-diazabicyclo[5.4.0]undec-7-ene at 20–25 °C afforded, after an acidic workup, the targeted molecule in good yield at a kilogram scale. Notable critical parameters and safety enhancements are defined and successfully implemented to produce three consecutive validation batches in a reproducible manner.
1. Introduction
Tirapazamine (SR259075, 1,2,4-benzotriazin-3-amine 1,4-dioxide, 1, Scheme 1) emerged as a lead compound that exhibits preferential toxicity toward hypoxic cells for one of our oncology programs. Tirapazamine 1 is a prodrug that is bioactivated only at very low levels of oxygen (hypoxia)1 by a one-electron bioreduction pathway, primarily by NADPH/cytochrome P450 reductase, to a highly toxic DNA-reactive radical.2
Scheme 1. Tirapazamine Structure.
Tirapazamine 1 underwent phase II/III testing in patients with head and neck, non-small-cell lung, cervical cancers and metastatic melanoma3 using an aqueous parenteral formulation.4 Process validation was therefore undertaken for tirapazamine 1 in anticipation of commercial annual supplies forecasted to be in the order of several dozen kilograms.
The process development work, scale-up, and the validation were conducted by scientists at Sanofi-Aventis. We describe herein the selection and development of the industrial validated synthesis of tirapazamine 1.
2. Results and Discussion
2.1. Process Selection
The literature precedents were reviewed to assist in identifying the best approach for the industrial development, as shown in Scheme 2. Mason and Tennant developed a two-step process.5 Reaction of 2-nitroaniline (2) with cyanamide afforded 1,2,4-benzotriazin-3-amine 1-oxide (3) in 80% isolated yield. Oxidation of 3 with 90% hydrogen peroxide in acetic acid yielded tirapazamine 1 in 63% yield. A second route was later developed by Suzuki and Kawakami involving a base-promoted heteroannulation of 2-fluoronitrobenzene (4) with guanidine for the synthesis of intermediate 3, followed by oxidation to tirapazamine 1.6 Although these routes are a simple two-step synthesis, the impurities generated were very difficult to remove and thus considered as a roadblock toward synthesis industrialization and validation. Seng and Ley described a one-step procedure from benzofuroxan-1-oxide 5 and disodium cyanamide and subsequently published a patent which reported a yield of 89%.7
Scheme 2. Overview of Main Processes for Tirapazamine 1 Synthesis.
Reagents and conditions: (a) H2NCN, neat, 100 °C, yield 80%; (b) H2O2 90% in AcOH, 50 °C, yield 63% or 60–70 °C, yield 28%; (c) guanidine, tBuOK, tetrahydrofuran, 50–60 °C, yield 82%; (d) Na2NCN, MeOH, H2O, 50–60 °C, then purification using HOAc, yield 89%
However, in our hands, an important drawback of Seng and Ley’s process was that the yields were variable at larger scales, varying from 17 to 39% on a 60 mol scale and required addition of solid disodium cyanamide. We, nevertheless, chose to focus on this one-step process, but the process was redesigned to obtain homogeneous conditions for scale-up.8
A full process safety study indicated that the average reaction temperature (∼70 °C) was far too close to the runaway reaction temperature of 85 °C. A differential reaction calorimeter showed that the addition of benzofuroxan 5 was highly exothermic, +70 °C, with a slight induction period before the main heat flow. Also, the use of dimethyl sulfoxide (DMSO), known to be unstable under a strong basic condition, was not acceptable.9 Additionally, cyanamide can decompose explosively if heated above 40 °C with an enthalpy of 1087 J/g. The process was considered very hazardous and therefore could not be scaled up.
2.2. Development of the Industrial Process
Our long-term goal was to validate the synthesis to produce tirapazamine 1 at the intended industrial scale. This implied the following:
technical scalability and good robustness of the processes;
good overall yield and process reproducibility (space–time–yield);
acceptable process safety and industrial hygiene impact.
The FDA Guidance10 defined the process validation as a collection and evaluation of data, from the process design stage through commercial production, which establishes scientific evidence that a process is capable of consistently delivering a quality product. In particular, the guidance describes process validation activities in three stages:
In stage 1, the commercial process is defined based on knowledge gained through development and scale-up activities,
In stage 2, the process is qualified; the process design is evaluated and assessed to determine if the process is capable of reproducible commercial manufacturing,
In stage 3, there is continued process verification during routine production.
If a reduction in the reaction temperature for Seng’s process were implemented in order to provide a better safety margin, it would extend the reaction time and allow the sodium salt of the drug substance to precipitate earlier in the reaction. The precipitated drug substance would cause a reduction in the heat-transfer rate and thereby further increase the contact time of the drug substance with the reaction mixture. It has been shown, by these authors, that tirapazamine 1 decomposes in an aqueous base. Seng’s process relied on the rapid addition of benzofuroxan 5 to minimize contact of the drug substance with the strongly basic reaction mixture.
In parallel, examples were found in the literature, indicating that benzofuroxan 5 reacts under mild nucleophilic conditions.11 Thus, we reasoned that benzofuroxan 5 should react with cyanamide under such mild conditions.
A few strong anhydrous bases were screened for use in the reaction instead of NaOH. 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU, 7) was found to facilitate tirapazamine 1 formation. The product could then be isolated after acidification with acetic acid.12 This process, described in Scheme 3, was selected for our industrial application.
Scheme 3. Tirapazamine Industrial Process.
Reagents and conditions: (a) CH3CN, 20–25 °C; (b) AcOH, then purification using CH3SO3H/water and AcONa/water
2.2.1. Establishing the DBU/Cyanamide/Benzofuroxan 5 Ratio
DBU is required as a base to deprotonate cyanamide and initiate the reaction with benzofuroxan 5. DBU also forms a salt with tirapazamine 1 in the reaction mixture. Therefore, 1 equiv is the minimum amount required in order to complete the reaction. Cyanamide is known to condense with itself under basic conditions to give dicyandiamide.13 A stability study was conducted to determine the amount of base necessary to maintain a stable solution of cyanamide in salt form. The results of this study are represented in Table 1.
Table 1. Stability Study of Cyanamide Salt Solution.
| equivalents of DBU used | remaining cyanamide day 1 | remaining cyanamide day 4 (%) | remaining cyanamide day 6 (%) |
|---|---|---|---|
| 1 | ∼80% | 50 | 30 |
| 1.5 | ∼80% | 70 | 40 |
| 2 | not available | 70 | 70 |
| 3 | not available | 70 | 70 |
This study showed that a minimum ratio of 2/1 DBU/NH2CN was necessary to obtain good stability for cyanamide. However, a better chemical yield of tirapazamine 1 was obtained when the ratio of DBU/NH2CN was 1/1 than when the ratio was 2/1. It is reasoned that too strong basic conditions disfavor the formation of tirapazamine 1. To quench the reaction, it was necessary to add enough acid to neutralize the amount of DBU used in the reaction. This is required to release the DBU/tirapazamine 1 salt and isolate the product. As the cyanamide and DBU ratio to benzofuroxan 5 increases, the reaction rate and the chemical yield of tirapazamine 1 also increase. A plateau effect on the chemical yield is observed for a ratio of 1/2/2 benzofuroxan 5/DBU/NH2CN, which gives almost the same conversion rate as a ratio 1/3/3 [ratio 1/3/3 (38% isolated yield) vs 42% for a ratio 1/2/2].
2.2.2. Establishing the Reaction Temperature
The upper limit of 40 °C was imposed for safety reasons. Reaction temperatures less than 20 °C had a slower reaction rate and resulted in poor conversion to tirapazamine 1 and higher decomposition of benzofuroxan 5. Therefore, the temperature was kept at the range of 20–25 °C.
2.2.3. Reaction Solvent
Acetonitrile was chosen as a reaction solvent as it was already defined as the best wash solvent in the reaction workup (see the Isolation and Purification Procedure section). The use of acetonitrile as a solvent improved the heat-transfer rate sufficiently so that the temperature was controlled with only a small temperature difference between the reactor jacket and the contents, provided that the addition was made progressively. The total acetonitrile present in the reaction mixture at the time was defined as 3.75 L per benzofuroxan 5 kg. Cyanamide is readily soluble in acetonitrile, even though its dissolution is endothermic. It was decided to add the acetonitrile for the reaction dilution along with the cyanamide. Therefore, acetonitrile was used to dissolve and add the cyanamide onto benzofuroxan 5 in the DBU solution.
2.2.4. Establishing the Reaction Time
The reaction was monitored quantitatively by high-performance liquid chromatography (HPLC), and a typical reaction course is depicted in Figure 1.
Figure 1.
Reaction kinetic study at a temperature between 20 and 25 °C.
According to this kinetic study, performed at a temperature between 20 and 25 °C, it was observed that the reaction reached a conversion plateau at about 48 h. After this mixing time, a significant amount of tirapazamine 1 was formed, whereas slow degradation of benzofuroxan 5 was observed. Therefore, the minimum reaction time was established at 48 h, as no increase of tirapazamine 1 proportion was observed later after this reaction time.
At full-scale reaction, the same kinetics were observed, as shown in Figure 2.
Figure 2.
Reaction kinetic study at temperatures between 20 and 25 °C.
The reaction reached a conversion plateau with a HPLC assay (weight formed of tirapazamine 1 vs the total reaction assay) unchanged after 60 h. An in-process control limit of >15 wt % tirapazamine 1 by the HLPC assay was established.
2.2.5. Effect of Water on the Reaction
Commercial grade benzofuroxan 5 is commonly supplied either wet with water (5–10%) or anhydrous. Therefore, the effect of water on the reaction outcome was investigated. Benzofuroxan 5, which contained 5% water, was utilized in the anhydrous reaction procedure. The corresponding purity was acceptable, but the yield, after correction for the water content in the benzofuroxan 5, was reduced by 2–4%. Then, reactions with benzofuroxan 5 containing a minimum amount of water were performed. The results were compared with the reaction with wet benzofuroxan. The results are shown in Table 2.
Table 2. Study of Water Contained in Benzofuroxan 5.
| benzofuroxan 5 used |
tirapazamine 1 produced | |
|---|---|---|
| supplier | water content (%) | yield (%) |
| internal | 0.3 | 38.7 |
| Aldrich | 0.23 | 36, 37 and 38 |
| Farchemia | 5 | 34.6 |
A specification of not more than 0.5% water was set for benzofuroxan 5 to ensure yield consistency. It is possible that water triggers stability issues under basic reaction conditions and therefore lowers the yield.
2.2.6. Isolation and Purification Procedure
The purification solvent was selected on a screening, represented in Table 3, with the purity assay observed after filtration of the crude tirapazamine 1, which was obtained according to the process described in the above sections.
Table 3. Purification Study.
| solvent | yield (%) | assay % |
|---|---|---|
| acetone | 42 | 93 |
| isopropyl alcohol | 51 | 87 |
| acetonitrile | 37–39 | >99 |
The addition of acetic acid is necessary for the neutralization of the DBU salt of tirapazamine 1. A 1% excess of acetic acid was used to ensure complete neutralization. The crystallization temperature is important in maximizing the yield. The crystallization temperature was set at 5–10 °C and the stirring time was set for 12–24 h to maximize the yield. Adequate washes of the crude tirapazamine 1 filtrate were found critical to ensure product quality. Then, the crude was dissolved in aqueous methanesulfonic acid with gentle warming (40–50 °C) until complete solution was obtained. The dissolution time was determined to require 3–6 h. Tirapazamine 1 was proven to be stable in aqueous methanesulfonic acid. Table 4 describes the summary of the main key parameters studied.
Table 4. Summary of Parameters Studied.
| parameter | value chosen |
|---|---|
| base | DBU |
| minimum DBU/tirapazamine 1 ratio | 1/1 |
| DBU/cyanamide ratio | 1/1 |
| DBU/cyanamide/benzofuroxan 5 ratio | 2/2/1 |
| minimum reaction time | 48 h |
| reaction temperature | 20–25 °C |
| water content | 0.5% water was set for benzofuroxan 5 |
| reaction solvent | acetonitrile, 3.75 L per benzofuroxan 5 kg |
Using these typical parameters, at the laboratory scale, the conversion rate of benzofuroxan 5 was 66% (internal standard DBU) and the chemical yield was 57% (internal standard DBU). The isolated yield was 38% after purification [assay: 99.2%, total chromatographic impurity: 0.29%].
2.3. Safety Precautions for Process Implementation
The synthesis was considered less hazardous than Seng and Ley’s process in DMSO particularly with regard to the reduction of the reaction temperature from 70 to 20–25 °C, the reason being that the reaction no longer took place at a temperature close to decomposition (85 °C) where the runaway risk was assessed to be very significant. However, it remained not intrinsically safe because of heat accumulation. Thus, introduction of the cyanamide solution in acetonitrile to the benzofuroxan 5 solution in DBU was carefully studied with regard to its exothermic behavior using a differential reaction calorimeter, as represented in Figure 3. Specific heat at the end of the reaction was measured to be 1.67 J·g–1·°C–1, corresponding to a thermal result of ΔQ = 318 J/g of benzofuroxan 5. Indeed, in case of loss of cooling, the heat of the reaction can increase the batch temperature to 46 °C. Moreover, the reaction lasts 50 h after a 30 min dosing. Therefore, the accumulation of heat in the reactor is significant.
Figure 3.
Exothermic profile of cyanamide addition on benzofuroxan 5.
A scaled down laboratory reaction was conducted to study the potential for a reaction quench in the event of a cooling failure during a large-scale reaction. Thus, right after cyanamide addition, water was added as rapidly as possible. The temperature, without external cooling, did not rise above 34/35 °C, which was below the accepted maximum safe operating temperature of 40 °C and thus considered safe enough to allow process implementation (Figure 4).
Figure 4.
Emergency quench of the reaction mixture with water.
2.4. Process Validation
Process validation was conducted for commercial manufacturing of a tirapazamine 1 drug substance on a 10 kg scale of benzofuroxan 5, which was purchased from an external supplier. The following critical parameters were included in the validation activity: batch temperature (addition of cyanamide solution to benzofuroxan 5 in DBU) 5–25 °C; batch reaction temperature 20–25 °C for 60–72 h; crystallization temperature of the crude tirapazamine 1 5–10 °C for 12–24 h; absorbance in the washes (NMT 0.08%) and DBU content (NMT 0.2%), reactor temperature for solubilization of crude tirapazamine 1 in aqueous methanesulfonic acid, 40–50 °C for 3–6 h; sodium content of the water washes of purified tirapazamine 1, NMT 1200 ppm; drying temperature at 50–55 °C for 24 h; screening material through a 25 mesh, particle <900 μm. Table 5 summarizes the overall validation results for the three consecutive batches.
Table 5. Summary of Validation Results.
| parameter | 1002-H | 1003-H | 1004-H |
|---|---|---|---|
| batch temperature | 22–24 °C | 21–28, 3 °Ca | 22 °C |
| reaction time | 72 h | 66 h | 66 h |
| yield | 34.9% | 35.5% | 35.4% |
| mass | 4.3 kg | 4.4 kg | 4.4 kg |
| analytical results | compliant | compliant | compliant |
Temperature excursion on batch 1003-H was assessed to be a minor deviation caused by a failure of a temperature control module.
The validation protocol conditions for the synthesis were met by the three validation batches, yields were considered reproducible, and reaction conditions robust. The recoverability of the waste streams, DBU or acetonitrile, was not investigated because of the small annual volume anticipated.
3. Conclusions
The synthesis procedure has been improved from the original heterogeneous reaction that required addition of a solid starting material with an uncontrolled exotherm. The final industrial process addressed the exotherm and process impurities. It was possible to implement it safely with appropriate precautions. It has been demonstrated that this process is reproducible and well controlled and produces a good quality material. This process has been validated at the intended commercial scale. FDA guidance stages 1 and 2, regarding process design and validation, were closely followed, although stage 3 was not possible as tirapazamine 1 was not commercially launched.
4. Detailed Experimental Section
Tirapazamine, 1,2,4-benzotriazin-3-amine 1,4-dioxide (SR259075) 1: A stirred solution of cyanamide (5.8 kg, 137.96 mol, 2.16 equiv) in acetonitrile (27 kg) at 15–20 °C was added to a stirred suspension of benzofuroxan 5 (9.5 kg, 69.79 mol) and DBU (21.3 kg, 139.88, 2 equiv) at 5–10 °C for over about 100 min while maintaining the reaction temperature between 15 and 25 °C. The system was diluted with 1 kg of acetonitrile, and stirring was maintained for about 65 h at 20–25 °C. The slurry was charged with 45 kg of acetonitrile, cooled to 5–10 °C, and charged with 8.5 kg of acetic acid. The system was rinsed and diluted with 3 kg of acetonitrile. The mixture was cooled to 0–10 °C and stirred for an additional 46 h. The mixture was filtered, and the solid was washed with 3 × 15 kg of acetonitrile. The solid was collected and slurried twice with 20 kg of water and washed again with 10 kg of water.
Purification: The crude solid tirapazamine 1 salt was charged to a reactor with 29 kg of purified water. The slurry was stirred and charged with 8.7 kg of methanesulfonic acid and heated to 40–50 °C to obtain a solution. The solution was filtered and transferred onto a solution of 9.2 kg sodium acetate in 49 kg of purified water to precipitate the tirapazamine 1. The suspension was cooled to 10–25 °C and stirred for at least 1 h. The suspension was filtered, and the solid was washed with 10 kg of purified water, resuspended in 20 kg of purified water, filtered, and then washed again with 10 kg of purified water. The solid was slurried with 20 kg of purified water, filtered, and washed with 10 kg of purified water. The solid was dried in a vacuum dryer at 50–55 °C for 24 h, and the dry solid tirapazamine 1 was passed through a 25-mesh screen to obtain a material with particles <900 μm.
1H NMR (DMSO-d6, 500 MHz): δ ppm: 7.577 (d, 1H, J = 8 Hz), 7.946 (d, 1H, J = 8 Hz), 8.028 (2H, brs), 8.151 (t, 1H, J = 8 Hz), 8.212 (t, 1H, J = 8 Hz). 13C NMR (DMSO-d6, 125 MHz): δ 117.5, 121.6, 127.3, 131.1, 135.8, 138.9, 151.9. HRMS: exact mass (by Xevo QToF), MH+; found, 179.0559 (MH+ calcd 179.0569, difference: 5.6 ppm). Elementary analysis: found C 46.87%, H 3.30%, N 30.93%. Calcd C, 46.57%; H, 3.50%; N, 31.03%.
Acknowledgments
This work is published 10 years after the original validation activities. We are grateful to the many colleagues within Sanofi who provided safety assessments and analytical support including a special thanks to Frederic Herman for all spectroscopic analysis. In particular, this paper is dedicated to all colleagues of the Sanofi Great-Valley (PA) and Organichem facility in Albany (NY).
Supporting Information Available
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.0c01250.
1H NMR, 13C NMR, MS, and IR spectra for the synthesized compound, tirapazamine 1 (PDF)
The authors declare no competing financial interest.
Supplementary Material
References
- a Jounaidi Y.; Waxman D. J. Combination of the bioreductive drug tirapazamine with the chemotherapeutic prodrug cyclophosphamide for P450/P450-reductase-based cancer gene therapy. Cancer Res. 2000, 60, 3761–3769. [PubMed] [Google Scholar]
- a Lloyd R. V.; Duling D. R.; Rumyantseva G. V.; Mason R. P.; Bridson P. K. Microsomal reduction of 3-amino-1, 2, 4-benzotriazine 1, 4-dioxide to a free radical. Mol. Pharmacol. 1991, 40, 440–445. [PubMed] [Google Scholar]; b Fitzsimmons S. A.; Lewis A. D.; Riley R. J.; Workman P. Reduction of 3-amino-l, 2, 4-benzotriazine-l, 4-di-N-oxide (tirapazamine, WIN 59075, SR 4233) to a DNA-damaging species: a direct role for NADPH: cytochrome P450 oxidoreductase+. Carcinogenesis 1994, 15, 1503–1510. 10.1093/carcin/15.8.1503. [DOI] [PubMed] [Google Scholar]; c Patterson A.; Barham H.; Chinje E.; Adams G.; Harris A.; Stratford I. Importance of P450 reductase activity in determining sensitivity of breast tumour cells to the bioreductive drug, tirapazamine (SR 4233). Br. J. Canc. 1995, 72, 1144–1150. 10.1038/bjc.1995.478. [DOI] [PMC free article] [PubMed] [Google Scholar]; d Anderson R. F.; Shinde S. S.; Hay M. P.; Gamage S. A.; Denny W. A. Activation of 3-amino-1, 2, 4-benzotriazine 1, 4-dioxide antitumor agents to oxidizing species following their one-electron reduction. J. Am. Chem. Soc. 2003, 125, 748–756. 10.1021/ja0209363. [DOI] [PubMed] [Google Scholar]
- a Hicks K. O.; Myint H.; Patterson A. V.; Pruijn F. B.; Siim B. G.; Patel K.; Wilson W. R. Oxygen dependence and extravascular transport of hypoxia-activated prodrugs: comparison of the dinitrobenzamide mustard PR-104A and tirapazamine. Int. J. Radiat. Oncol. Biol. Phys. 2007, 69, 560–571. 10.1016/j.ijrobp.2007.05.049. [DOI] [PubMed] [Google Scholar]; b Cowen R. L.; Williams K. J.; Chinje E. C.; Jaffar M.; Sheppard F. C. D.; Telfer B. A.; Wind N. S.; Stratford I. J. Hypoxia targeted gene therapy to increase the efficacy of tirapazamine as an adjuvant to radiotherapy. Cancer Res. 2004, 64, 1396–1402. 10.1158/0008-5472.can-03-2698. [DOI] [PubMed] [Google Scholar]; c Reddy S. B.; Williamson S. K. Tirapazamine: a novel agent targeting hypoxic tumor cells. Expet Opin. Invest. Drugs 2009, 18, 77–87. 10.1517/13543780802567250. [DOI] [PubMed] [Google Scholar]
- Brown S.; Baker E. U.S. Patent 5,827,850 A; U.S. Patent and Trademark Office: Washington, DC, 1998.
- Mason J. C.; Tennant G. Heterocyclic N-oxides. Part VI. Synthesis and nuclear magnetic resonance spectra of 3-aminobenzo-1, 2, 4-triazines and their mono-and di-N-oxides. J. Chem. Soc. B 1970, 911–916. 10.1039/j29700000911. [DOI] [Google Scholar]
- a Suzuki H.; Kawakami T. A convenient synthesis of 3-amino-1, 2, 4-benzotriazine 1, 4-dioxide (SR 4233) and related compounds via nucleophilic aromatic substitution between nitroarenes and guanidine base. Synthesis 1997, 1997, 855–857. 10.1055/s-1997-1283. [DOI] [Google Scholar]; b Moskalev N. V.; Gribble G. W. U.S. Patent 7,129,349 B2; U.S. Patent and Trademark Office: Washington, DC, 2006.
- a Seng F.; Ley K. Einfache Synthese von 3-Aminobenzo-1,2,4 triazin-l,4dioxid. Angew. Chem. 1972, 84, 1061. 10.1002/ange.19720842109. [DOI] [Google Scholar]; b Ley K.; Seng F. Synthesen unter verwendung von benzofuroxan. Synthesis 1975, 1975, 415–422. 10.1055/s-1975-23779. [DOI] [Google Scholar]; c Ley K.; Seng F.; Metzger K. G. U.S. Patent 3,868,371 A, 1975.
- Philion R. E. U.S. Patent 5,672,702 A, 1995.
- Bretherick L.Bretherick’s Handbook of Reactive Chemical Hazards, 7th ed; Elsevier, 2018; p 381. [Google Scholar]
- US Food and Drug Administration . Guidance for industry. Process validation: General principles and practices; US Department of Health and Human Services, Food and Drug Administration, 2011. [Google Scholar]
- Borah H. N.; Sandhu J. S. Base catalysed reaction of 2-cyanomethyl-1, 3-benzothiazole with benzofuroxanes. Heterocycles 1986, 24, 979–982. 10.3987/r-1986-04-0979. [DOI] [Google Scholar]
- Burgos A. U.S. Patent 6,953,852 B2, 2005.
- Miller S. A.; Bann B. Cyanamide. J. Appl. Chem. 1956, 6, 89–93. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.