148d

DEVELOPMENT OF AN ENVIRONMENTALLY FRIENDLY,

COST EFFECTIVE PROCESS FOR
THE PRODUCTION OF CYTOVENE® ANTIVIRAL AGENT

Gary B. Semones, Ph.D., P.E., Sam L. Nguyen, Ph.D., J.D.,

Yeun-Kwei Han, Ph.D., Eric Lodewijk, Ph.D., and George Schloemer, Ph.D.

Boulder Technology Center

Roche Colorado Corporation
2075 North 55th Street
Boulder, CO 80301-2880
[Link]@[Link]
[Link]@[Link]

Key words: Cytovene, Ganciclovir, batch process modeling, Responsible Care.

Prepared for Presentation at the 2000 Spring AIChE Meeting, Atlanta, GA March 5-10.

Copyright © Gary Semones, Sam L. Nguyen, Yeun-Kwei Han,

Eric Lodewijk, and George Schloemer
1/2000, Roche Colorado Corporation
Unpublished

AIChE shall not be responsible for statements or opinions contained in papers or printed in its publications.
 148d

Introduction

One of the missions of the Roche Colorado Corporation’s Boulder Technology

Center is to develop second generation processes for the production of pharmaceutical

compounds. The Tech Center, as it is known, develops new chemistry and implements

state-of-the-art engineering technology to create less expensive processes, reduce cycle

times, improve throughput, and minimize the impact to the environment.

Roche Colorado Corporation is a Responsible Care company. The Chemical

Manufacturers Association (CMA) launched the Responsible Care program in 1988 to

respond to public concerns about the manufacture and use of chemicals. Part of the

guiding principles of the Responsible Care program is to make health, safety, the

environment, and resource conservation critical considerations for all new and existing

products and processes. Scientists and engineers at the Tech Center perform second

generation process development in accordance with the guiding principles of the CMA’s

Responsible Care program.

This paper discusses the development of a novel environmentally friendly process

(i.e., “Green chemistry”) for the production of Cytovene (ganciclovir), a

pharmaceutical for the treatment of cytomegalovirus (CMV) retinitis in

immunocompromised patients. The second generation Guanine TriEster (GTE) Process

reduced air emissions by ∼66% and liquid/solid waste generation by ∼89%. The

increases in raw material costs were more than offset by improvements in equipment

utilization as reflected in the overall process throughput. Thus the objectives of the

Responsible Care program were met while also meeting and exceeding the profitability

and production capacity demands of the corporation.

2
 148d

Background

Process research and development activities toward establishing second

generation processes for the reduction of manufacturing costs for bulk pharmaceutical

drug substances has been a central goal for process chemists and engineers at the Roche

Boulder Technology Center. The synergistic application of the basic principles of

process modification through the development of novel reaction pathways, in conjunction

with efficient chemical engineering principles have been realized in the successful

demonstration of the new Guanine TriEster (GTE) Process for the manufacture of

Cytovene®.

Cytovene® is a potent antiviral agent for the treatment of cytomegalovirus

(CMV) retinitis infections in immunocompromised patients, including patients with

AIDS. Patients who are particularly at risk for developing CMV infections include those

with AIDS and patients who are recipients of solid tissue transplants.

HN N

H2N N N
OH
O
OH
Ganciclovir
(CYTOVENE)

3
 148d

This presentation provides an ex post facto detailed analysis and comparison of

the relative manufacturing efficiencies between the first generation Persilylation Process

and the second generation GTE Process for the commercial manufacture of ganciclovir

using a commercial modeling software program (Batch Plus from Aspen Technology,

Inc.). Specific parameters calculated for documenting the relative efficiencies of the two

processes include estimated costs, energy usage, mass and energy balances, estimated

cycle times, air emissions, and solid and liquid waste streams. The qualitative results of

this evaluation revealed that significant process efficiencies resulting in cost reduction

and reduced global environmental impact can be attained from the judicious chemical

modification to a convergent synthesis, the selective application of the principles of

molecular conservation, and application of creative engineering principles for optimal

process design.

In 1974, scientists at Wellcome discovered the potent antiviral agent acyclovir

(Zovirax®) for the treatment of various viral infections including herpes viruses HSV-1

and HSV-2. The corresponding bis-hydroxy homologue, generically known as

ganciclovir (Cytovene®), which was developed by Syntex, proved to be more effective in

inhibiting cellular DNA polymerase that is associated with viral infections. Subsequent

to these early discoveries, significant advances have been made to prepare various

guanosine analogs for the inhibition of viral activities related to both RNA and DNA

viruses (Robins, 1986).

The major synthetic challenge encountered with the preparation of ganciclovir

and related guanosine derivatives involve the selective N-alkylation of guanine and its

derivatives. In particular, the selective alkylation reaction of guanine affording the

4
 148d

desired N-9 product over the N-7 isomer is an exceptional challenge. This is due in part

to the lack of stereoelectronic differentiation between the N-9 and the N-7 centers and the

known kinetic isomerization and thermodynamic equilibration of the corresponding

alkylated product mixtures (Singh et al., 1999). In addition, the amphoteric property and

the highly insoluble characteristics of guanine in most organic solvent systems limit the

application of the standard reaction conditions (Claussen and Juhl-Christensen, 1993).

Many different approaches have been explored for the selective functionalization

of either guanine or various guanine derivatives. Previously reported conditions for the

selective alkylation reaction, that is, N-alkylation processes that afford predominantly or

exclusively the desired N-9 over the undesired N-7 isomer, include the use of reaction

conditions not ordinarily considered to be favored by environmentalists.

In particular, the application of coupling mediators comprising in part of cesium

carbonate (CsCO3) (Kim et al., 1988) and cesium iodide (CsI) (Matsumoto et al., 1970) in

chlorinated hydrocarbons such as dichloromethane, and toxic metal salts such as mercuric

acetate (Hg(OAc)2) (Hrebabecky and Farkas, 1974), mercuric cyanide (Hg(CN)2) (Kim et

al., 1991), zinc chloride (ZnCl2) (Matsumoto et al., 1970) and tin tetrachloride (SnCl4)

(Garner and Ramakanth, 1988). Incidental reaction conditions employed with the intent

to increase the solubility with concomitant increase in selectivity of alkylation involving

substituted guanine derivatives have included harsh alkylating conditions such as

dimethylformamide (Slusarchyk et al., 1989), dimethylsulfoxide (Phadtare and Zemlicka,

1989), sulfolane (Jacobs et al, 1989), and related high boiling solvent systems.

The first commercially viable process (Verheyden et al., 1986) for the

manufacture of ganciclovir was developed by Roche Colorado Corporation, formerly

5
 148d

known as Syntex Chemicals (see Scheme 1, Persilylation Process). The six-step process

required the processing of 28 reagents and intermediates, including the purification and

isolation of 5 discrete intermediates. Overall, the process afforded specification grade

ganciclovir in 54% yield.

Development of the GTE Process

By early 1992, projected commercial demand for ganciclovir reached

approximately 50 metric tons per year, and it was readily determined that the first

generation Persilylation Process could not be used to meet these production quantities

without a significant amount of capital investment in Roche’s production facilities. In

1993, the Boulder Technology Center completed the demonstration of a new and

expedient process for the production of ganciclovir by 1) leveraging the basic principles

of molecular conservation to minimize the creation and disposal of undesired wastes, and

2) formulating efficient process engineering design for streamlining process operation

and the recycling of raw materials.

The second generation Guanine TriEster (GTE) Process (Scheme 2: The GTE

Process) successfully demonstrated the potential for a “one step” process for the

production of ganciclovir, reduced the number of chemical reagents and intermediates

from 28 to 11, eliminated the two hazardous solid waste streams, and eliminated 11

different by-products from the liquid waste streams (Lodewijk et al., 1996). In addition,

of the 5 raw materials employed but not incorporated into the molecular structure of the

final product, 4 were efficiently recovered in situ and reused for subsequent production

batches.

6
 148d

The chemistry scheme designed into the GTE Process also eliminated a

potentially hazardous palladium catalyzed hydrogenation step, a procedure previously

mandated by the nature of dibenzyl ether protecting group incorporated in the

Persilylation Process.

Furthermore, particularly critical to the success of applying the principles of

molecular conservation was the judicious design of the 4-carbon triester coupling reagent,

which was ultimately appended into the purine nucleus as the key N-9 substituted side

chain. The triester coupling reagent was designed to incorporate simple low molecular

weight functionalities, vis-a-vis an ethyl ester that, after condensation, would be readily

unmasked under simple hydrolytic conditions to provide a minimal amount of innocuous

by-products. In practice, the construction of the low molecular weight coupling reagent

was readily accomplished using glycerol and propionic acid derivatives.

7
 148d

O
OH OH
HN N

OH H2N N N
H
CH2(OMe)2
2 Bnzl-Cl Ac2O
Ac2O DMAP, PhCH3
p-TsOH
OBn OBn
O
+ MeOCH2OAc
N
OH HN

p-TsOH AcNH N N
H
Hexanes
HMDS, Xylenes
(NH4)2SO4, DMF
OBn OBn
1

1 O
AcO O OBn
O N
O HN
OBn
HN N AcNH N N

AcNH N N O
SiO2-AlO2 N-7 50% Yield
N-9
CH2Cl2 N-9/N-7=10:1
OBn OBn

WASTE H2
Pd(OH)2/MeOH

O O

N N
HN HN
NH4OH, H2O
N AcNH N N
H2N N

O MeOH O
Ganciclovir

OH OH OH OH

SCHEME 1: Persilylation Process

8
 148d

Another key aspect for attaining higher reaction yields in the GTE Process was

the novel design of the silylation reaction of guanine, a raw material prevalently found in

animals and plants and isolated from guano (Shapiro, 1968). Minor processing

modifications of the silylation reaction of guanine and subsequent coupling with the

triester reagent gave rise to a highly regioselective alkylation product, thereby

minimizing the formation of undesired by-products. Furthermore, the in situ acylation of

the N-9/N-7 product mixture followed by the selective crystallization of the N-9 product

provided a facile method for isolating exclusively the desired N-9 product in high

chemical yield.

Final deprotection to remove the esters and amide protecting groups could be

readily accomplished in a single step using aqueous ammonium hydroxide, to afford the

desired ganciclovir in 65% yield overall.

Practical engineering designs for further streamlining the GTE Process included

the in situ hydrolysis of the N-9 acylated intermediate thereby establishing a “one-step”

O
1. HMDS, TfOH
EtCO2 OCOEt HN N
O
O
N OCOEt H2N N N
HN

N O
H2N N (EtCO)2O, DMAP
H
MeOH, PhCH3
2. NH4OH, MeOH OH OH

Ganciclovir

Scheme 2: The Guanine TriEster (GTE) Process

9
 148d

synthesis of ganciclovir and eliminated the drying, isolation and storage of the triester

intermediate. Accordingly, the crystallized pure acylated N-9 product is filtered from the

reaction mixture, washed with toluene, and then the product in the filter is dissolved in

aqueous ammonium hydroxide as the toluene wet cake into a reaction vessel and

hydrolyzed in the same medium to afford ganciclovir.

When the ammonium hydroxide hydrolysis reaction is complete, gaseous

ammonia from the reaction mixture is recovered through a vacuum distillation utilizing a

secondary receiver containing cold water to absorb the ammonia. The innovative method

of recovering gaseous ammonia under these conditions permitted the removal of

ammonia gas from the product mixture with the concurrent purification of ammonium

hydroxide to be recycled in the subsequent production batches.

Based on the initial projected demand for the commercial production of 50 metric

tons of ganciclovir per year, the annual consumption of raw materials, waste streams and

processed quantity of reaction intermediates are tabulated in Table 1.

10
 148d

Table 1: Raw Material Usage, Waste Streams and Intermediates

ESTIMATED ANNUAL CONSUMPTION OR PROCESSED1

(Kg)
REAGENTS PERSILYLATION PROCESS GTE PROCESS
Guanine 17,800 17,800
HMDS 89,000 (>90% recycl) 89,000 (>90% recycl)
CF3SO3H 356
TriEster Reagent 48,060
(RCO)2O (R=Me, Et) 30,000 22,960
DMAP 1,780 1,167
MeOH 1,600,000 114,000 (>90% recycl)
Toluene 15,000 231,000 (>90% recycl)
NH4OH(aq) 900,000 110,000 (>70% recycl)
Dimethoxymethane 131,904
P-TsOH 2,400
Dichloromethane 95,000
Hexanes 125,400
Benzyl Chloride 40,000
Glycerol 10,720
Triethylamine 3,000
Pyridine 3,000
Xylenes 50,000
Dimethylformamide 3,100
Ammonium Sulfate 1,500
Silica Alumina 15,0002
Hydrogen gas 7,000
Palladium Hydroxide 3,2002
N-Acetyl guanine 19,094
Dibenzyl Glycerol 34,300
Methomethylacetate 190,000
Dibenzylglyceryl methylacetate 34,290
N-Acetyl DB-ganciclovir 96,364
N-Acylated ganciclovir (N-7)3 5,000 1,430
N-Acetyl ganciclovir 54,000
TOTAL PROCESSED 3,577,852 635,773
TOTAL PROCESSED AFTER 3,523,437 193,858
RECYCLE

1 Based on a 50 MT annual production of ganciclovir.

2 Solid waste streams.
3 Material in liquid waste streams.

11
 148d

At the 50 metric ton annual production level, a comparison of the processed raw

materials and intermediates between the two processes reveals that approximately 3.58

million kg are required in the Persilylation process while only 636 thousand kg are

required in the GTE process. Implicit in this disparity of material usage are the

significant global savings associated with the initial sourcing, production, shipping,

handling, warehousing, processing and disposal of ancillary raw materials, intermediates

and waste by-products. Significantly reduced environmental impact are concurrently

achieved in the decrease in associated energy consumption and emissions in the

processing of fewer raw materials (vide infra).

Comparison of the Persilylation and GTE Processes

For purposes of this paper, the Persilylation and GTE processes were compared

by means of a series of computer simulations using Aspen Technology, Inc. Batch Plus

software. While production data are available for running both processes in

manufacturing-scale equipment, the processes were run using different equipment

combinations, different utilities, and at different scales. These parameter variations could

potentially bias the process evaluation unfairly in favor of one process or the other. The

use of Batch Plus software facilitated the modeling of the two operations assuming

identical equipment and utilities, identical operations protocols, and identical scales.

Output from the models was compared to process data to evaluate the accuracy of the

models. Modifications were made as necessary to ensure the accuracy of the process

models. Where information was either unknown or not available, the models assumed

the default conditions in the modeling software. The results of the comparison of the

12
 148d

Persilylation and GTE processes are shown in Table 2. Discrepancies between entries in

Tables 1 and 2 are due to differences between forecast product and raw material demands

(Table 1) and the processes as modeled based on final recipes (Table 2).

Table 2: Comparison of the Persilylation and GTE Processes for the Production of Ganciclovir
(Basis: 1,000 kg Ganciclovir output)

Persilylation GTE ∆%
Process Process

Number of Raw Materials 20 9 -55.0%

Weight of raw materials (kg) 38,469 16,892 -56.1%

3
Total reactor volume required (m ) 58.0 27.2 -53.1%

Vapor emissions (kg) 2,102 712 -66.1%

Solid and liquid waste (kg) 39,990 4,315 -89.2%

Energy for heating (kcal) 3,746,110 1,985,926 -47.0%

Energy for cooling (kcal) 1,463,620 369,286 -74.8%

Overall process yield 54% 65% 11%

Cycle Time (h) 49 29 -40.8%

3
Throughput (kg/m -h) 0.35 1.27 262.9%

Fresh Solvent Utilization (kg) 18,742 1,133 -94.0%

As shown, both the number and weight of raw materials consumed to produce

1,000 kg of Cytovene is greatly reduced in the GTE Process versus the Persilylation

Process. This is the result of several factors. The overall chemistry of the GTE Process

is greatly simplified as compared to the Persilylation Process. In full-scale processing,

there is a total of only five reactions and the isolation of one intermediate in the synthesis

13
 148d

using the GTE Process versus eight reactions involving the isolation of four intermediates

in the Persilylation Process. Eliminating reactions and isolations reduces both the

number of reagents required and the amount of solvent consumed during filtration and

washing the filter cake. In addition to simplified process chemistry, the GTE Process has

an 11% increase in yield from guanine to Cytovene.

The optimized chemistry and elimination of process steps are also reflected in

reduced process emissions and decreased liquid/solid waste generation. This results from

decreased solvent handling (e.g., solvent strip operations, filtrations, drying), process

optimization to eliminate solid by-products and decrease the use of filter-aids, and also

the implementation of solvent recovery operations. The GTE Process recovers over 90%

of the input solvents. This is also reflected in the fresh solvent utilization entry in Table

2. Furthermore, the GTE Process requires 94% less fresh solvent that the Persilylation

process.

By eliminating process steps involving reactions, heat-ups, cool-downs, solvent

strips, and drying operations, the utility demand is significantly decreased for the GTE

Process. The heating demand is down 47% and the cooling demand decreased by ∼75%.

This represents not only a cost savings for production, but decreases fossil fuel demands

and the corresponding emissions resulting from energy production.

Raw materials for the GTE Process were about 2.5 times as expensive as in the

Persilylation Process. However, this increase was offset by several factors including

those discussed above. The factor that made it financially attractive to pursue the second

generation chemistry was the increase in process throughput. This is an indicator of the

efficiency of equipment and time utilization and is measured in mass of product produced

14
 148d

per cubic meter of total vessel volume per hour of process cycle time. For the GTE

Process the throughput was 1.27 kg/m3-h versus 0.35 kg/m3-h for the Persilylation

Process – an increase of approximately 263%.

In addition to the environmental benefits and cost effectiveness of implementing

the second generation GTE Process for Cytovene, there are secondary, less quantifiable

benefits as well. By streamlining operations in the GTE Process, the isolation and

storage of several process intermediates were eliminated as well. This decreases the

“Goods in Process” inventory and frees valuable warehouse space that could be used for

other demands or be eliminated entirely.

Another benefit is a reduction in burden on the environment resulting from

decreased raw material consumption. The more efficient use of raw materials and

solvents decreases the demand on natural resources, reduces the consumption of fossil

fuels, and decreases emissions and waste streams produced by suppliers of raw materials.

Conclusion

The second generation Guanine TriEster (GTE) Process for the manufacture of

ganciclovir, a potent antiviral agent for the treatment of cytomegalovirus (CMV) retinitis

infections, was successfully developed at Roche Colorado Corporation to meet the

commercial demand of the market. By leveraging the synergistic collaborative efforts of

developing new and convergent chemical pathways with creative engineering principles,

the GTE Process permitted Roche to meet the significantly increased commercial demand

for ganciclovir and, at the same time, significantly reduced the potentially significant

environmental impact associated with large scale manufacture. By using a commercial

15
 148d

process modeling program, an accurate assessment of the qualitative efficiencies between

the first generation Persilylation and GTE processes may be determined. Second

generation pharmaceutical process development resulting in environmentally responsible

processes is indicative of Roche Colorado Corporation’s commitment to the environment

and to the Chemical Manufacturers Association Responsible Care program.

Acknowledgements:

The initial drive and creativity of the original co-inventors of the GTE Process

described in U.S. patent 5,565,565, including Drs. Eric Lodewijk, Yeun-Kwei Han, and

George Schloemer are gratefully acknowledged. Implementation of the GTE Process for

commercial production would not have been possible without the full support of our

friends and colleagues at the Boulder Technology Center and Roche Colorado

Corporation. This work was supported by Roche Colorado Corporation, a subsidiary of

Roche Holding Ltd.

16
 148d

References

Batch Plus Batch process modeling software from Aspen Technology, Inc. Ten Canal

Park, Cambridge, Massachusetts 02141-2200.

Claussen, F. P.; Juhl-Christensen, J. Organic Prep. & Proced. Intl. 1993, 25, 375 and

references cited therein.

Garner, P.; Ramakanth, S. J. Org. Chem., 1988, 53, 1294.

Hrebabecky, H.; Farkas, J. Coll. Czech. Chem. Comm., 1974, 39, 2115.

Jacobs, G. A.; Tino, J. A.; Zahler, R. Tetrahedron Lett., 1989, 30, 6955.

Kim, C. U.; Misco, P. F.; Luh, B. Y.; Hitchcock, M. J. M.; Ghazzouli, I.; Martin, J. C. J.

Med. Chem., 1991, 34, 2286.

Kim, Y. H.; Kim, J. Y.; Lee, C. H. Chemistry Lett. 1988, 1045.

Lodewijk, E.; Han, Y-K; Schloemer, G. C.; Nguyen, S. L. Preparation of N-9 Substituted

Guanine Compounds. U.S. Patent 5, 565, 565, October 15, 1996.

Matsumoto, H.; Kaneko, C.; Yamada, K.; Takeuchi, T.; Mori, T.; Mizuno, Y. Chem.

Pharm. Bull. Jpn, 1970, 18, 172.

Phadtare, S.; Zemlicka, J. J. Am. Chem. Soc., 1989, 111, 5925.

Reitz, A. B.; Rebarchak, M. C. Nucleosides & Nucleotides, 1992, 11, 1115.

Robins, R. K. in “Synthetic Antiviral Agents” in Chemical and Engineering News, Special

Report, 1986, 28-40.

Shapiro, Progr. Nucleic Acid Res. Mol. Biol. 1968, 8, 73-112. For a recent preparation,

see Chen, X.; Wang, Y. Chinese Journal of Pharmaceuticals, 1992, 23, 32-33.

Singh, D.; Wani, M. J.; Kumar, A. J. Org. Chem. 1999, 64, 4665-4668.

17
 148d

Slusarchyk, W. A.; Young, M. G.; Bisacchi, G. S.; Hockstein, D. R.; Zahler, R.

Tetrahedron Lett., 1989, 30, 6453.

Verheyden, J. P. H.; Martin, J. C.; McGee, D. P. C. Process for Preparing 2,6-

Substituted-9-(1,3-dihydroxy-2-propoxymethyl)-Purines and Certain Derivatives.

U.S. Patent 4, 621,140, November 4, 1986.

Verheyden, J. P. H.; Martin, J. C. Process for Preparing Guanine Derivatives. U.S. Patent

4,803,271, February 7, 1989.

Morgan, D. J.; Chapman, H. H. Alkoxy Methyl Ether and Alkoxy Methyl Ester

Derivatives. U.S. Patent 5, 225, 590, July 6, 1993.

18