CHEMTECH

November 1996

CHEMTECH 1996, 26(11), 28-33.
Copyright © 1996 by the American Chemical Society.

Pergolide:
Process design challenges of a potent drug

A multistep process that was hazardous to operating personnel has been replaced by one that requires no handling of dangerous dried solid intermediates. The quaternization/demethylation method developed for this process may have more general applicability.

by Jerry W. Misner, Joseph H. Kennedy, and W. Scott Biggs

Pergolide is one member of a class of potent dopamine receptor agonists used in the treatment of Parkinson's disease (see sidebar). Initially, pergolide was synthesized via several variations of a multiple-step process (4-6) consisting of up to six chemical transformations, a salt formation, and associated recrystallizations (Figure 1). In a pilot-plant setting, the process required 4-6 months to complete and resulted in overall product yields of 15-23%. Each intermediate was isolated, dried, evaluated for purity, and recrystallized -- if necessary -- before additional processing. Consequently, the operating personnel had to handle dried solids at least 18 times per manufacturing campaign.

TO SIDEBAR: Parkinson's disease and its treatment

Employee contact was a serious safety concern during drug production because the transfer of these solids produced hazardous dust. Worker protection consisted of several changes of protective clothing, cumbersome full-containment suits with airline respirators, dry-bag transfer apparatus, walk-in hoods, and negative air pressure maintenance for the limited-access manufacturing area. Dust monitoring and decontamination procedures added significantly to the complexity of the process. The search for a more streamlined process that would allow greater throughput and higher yields as well as minimize the handling of dry solids (thus enhancing worker safety) became a priority.

Looking for alternatives

Our search for a new approach was bounded by some restrictions. We didn't consider a total synthesis from simple raw materials a viable option because of the complexity of assembling the ergoline skeleton (7). Dihydrolysergol, the starting material used in the synthesis, is expensive but can be purchased from several commercial sources. Ultimately, it is derived via the fermentation of Claviceps purpurea, a fungus parasitic to grasses and cereals (8), to produce either elymoclavine or ergotamine. Elymoclavine can be hydrogenated directly with Raney nickel to give dihydroelymoclavine, whereas ergotamine can be processed via several steps to dihydrolysergol (Figure 2). Dihydroelymoclavine and dihydrolysergol are identical compounds usually known by the latter name.

To minimize the hazards from handling dry solids, a new route would have fewer steps or would eliminate the drying requirements. The latter option would be difficult to implement because the usual isolation workup of ergolines involves product precipitation by the addition of water, which would be incompatible with most subsequent steps. A one-pot synthesis would be the most desirable solution to the handling issues.

The transformations in the conversion of dihydrolysergol to pergolide include replacing the methyl group at N-6 with a propyl and the hydroxymethyl functionality at C-8 with a thiomethyl ether. The lengthier of these two conversions is the alkyl exchange at N-6. Hutchins and Dux pointed the way toward a solution in their work (9). They disclosed an exceptionally selective procedure for the SN2 demethylation of quaternary ammonium salts by lithium thiopropoxide in hexamethylphosphoric triamide (HMPA). Other methods lacked the exquisite sensitivity, high yields, and mild conditions of their method. For example, the best of the other procedures used thiophenoxide anion in 2-butanone to produce a ratio of demethylation to deethylation of only 3.5:1 (10).

Figure 1. The multiple-step synthesis of pergolide mesylate. The synthesis required up to six chemical transformations, took 4-6 months to complete, and yielded only 15-23% product.










Figure 2. The starting material for the one-pot synthesis can be derived from products of ergot fermentation.







The virtues of the Hutchins and Dux method are attributable to the high reactivity of the nucleophile and to the polarity of the solvent that facilitate the reaction. Relative rates of SN2 reaction substrates are well known to be benzyl > allyl > methyl > ethyl > propyl, etc., and the rate difference reported for methyl versus ethyl was approximately 30:1 (11). Hutchins and Dux (9) reported selectivities of 23-100:1 for methyl versus ethyl and yields > 90% for all the reactions they tested.

The drawbacks of this dealkylation procedure are the specific requirements of lithium thiopropoxide as the nucleophile and the carcinogenic HMPA as solvent. We believed that thiomethoxide anion could replace thiopropoxide, especially because we had the relative rate advantage of methyl versus propyl and saturated ring alkyl substituents at N-6. We hoped that other polar solvents could substitute for HMPA.

To implement this selective demethylation strategy, we had to first quaternize N-6 of dihydrolysergol with a propyl substituent. The use of 1-iodopropane (propyl iodide) proved superior to that of other propyl halides, mesylates, and tosylates. However, commonly used solvents for this procedure -- such as chloroform and N,N-dimethylformamide (DMF) -- did not support even a 75% conversion to the ammonium iodide. Eventually, three satisfactory solvents were found: 1,3-dimethyl-3,4,5,6-tetrahydro-2(1H)-pyrimidinone (DMPU), 1,2-dimethyl-2-imidazolidinone (DMEU), and 1-methyl-2-pyrrolidinone (NMP). DMPU is well known as a replacement solvent for HMPA (12).

All three solvents facilitated the quantitative quaternization of dihydrolysergol to 5 (Figure 3, step1a) in 3-5 h at 75-80 °C in the presence of an acid scavenger such as sodium bicarbonate. The scavenger was necessary to neutralize small quantities of hydrogen iodide formed by elimination from propyl iodide. We selected NMP as our preferred solvent on the basis of cost and bulk availability.

Intermediate 5 and all the other quaternized ergolines we examined lacked the hydrophobic property of their parent compounds. Isolation was possible only via chromatography as an amorphous solid, not as a simple precipitation by adding water. Without the ability to readily isolate and purify intermediates, we were forced to pursue a strategy of multiple steps in a single reactor, which we preferred.

Although our variation of the Hutchins and Dux chemistry could be attempted at this point to convert 5 into intermediate 3 as used in the multiple-step process, we decided to investigate the more desirable possibility of installing a leaving group (preferably a mesylate or tosylate) and then attempting the simultaneous thiomethyl ether formation and demethylation with an excess of thiomethoxide. The preparation of 6 required the use of pyridine (or related bases picoline or lutidine) as both the base and the primary solvent, which meant diluting the NMP solution of 5 with several volumes of pyridine. Because the quaternization step worked best when run at high concentration, the substantial increase in volume for the mesylation (and subsequent steps) was not an insurmountable hurdle. Unlike the quaternization, we never achieved a quantitative conversion. As measured by HPLC, ~2% of unreacted 5 was typically detected, indicating that any residual 5 would be converted primarily to 3 as a byproduct if the selective demethylation worked as anticipated.

In contrast to 5, mesylate 6 proved to be unstable in solution, especially at temperatures >15 °C. However, it could be isolated as a stable solid via chromatography. Mass spectrometric evidence suggested the presence of the iodo and chloro derivatives of 6 in minor quantities (Figure 4). This finding was not a problem because all three compounds supplied acceptable leaving groups to nucleophilic attack by thiomethoxide and would convert to the identical product.

Figure 3. The one-pot synthesis of pergolide.












Figure 4. Halide derivatives are potential byproducts. These two derivatives of 6 create no problem in the synthesis.



We completed the synthesis of pergolide by combining a <0°C solution of 6 with an excess of a cold solution of sodium thiomethoxide in NMP. Analysis by HPLC revealed the very rapid formation of what proved to be transient intermediate 7 followed by slower demethylation to pergolide. Heating to 80°C accelerated demethylation. Isolation was readily achieved by addition of water to precipitate the solid and then filtration. After drying, the crude pergolide mass was typically 90-95% of the theoretical amount with an HPLC purity of 94-95%, representing an overall yield of 86-88% for the one-pot synthesis.

To achieve effective demethylation and thiomethyl ether formation, a minimum of 8 equiv sodium thiomethoxide per mole of intermediate 6 was required. In contrast to the specific conditions that Hutchins and Dux required (9), we found that sodium, potassium, and lithium thiomethoxide were equally effective. Furthermore, the various ratios and types of solvents used throughout the process had no effect on the outcome.

Initially, we generated our sodium thiomethoxide solution with an excess of methanethiol to sodium methoxide in NMP (or DMF, DMPU, etc.). However, because water did not have a deleterious effect on the chemistry and was subsequently useful in the isolation, we substituted sodium hydroxide solution for sodium methoxide. In both cases, it was important that excess methanethiol be used to ensure that all of the methoxide or hydroxide ion was consumed. We were not successful in implementing this process with commercially available sodium thiomethoxide unless excess methanethiol was added to it first, because the sodium thiomethoxide was contaminated with varying amounts of sodium hydroxide.

Byproducts create problems

There were 5-6% related substances as byproducts; these three (1-2% each) accounted for the majority of them.



As mentioned above, compound 3 was expected and found. Similarly, as anticipated with a selective demethylation, a small amount of depropylation was found, producing byproduct 8, 6-methylpergolide. The third byproduct (9) was unexpected. Independent synthesis of this homologated thioether of pergolide was unsuccessful, but it has been isolated and characterized from a chromatography fraction. The presence of a much smaller amount of double homologue was detected by mass spectrometry and HPLC. The mechanism for the formation of 9 remains uncertain, but we discovered that the use of <8 equiv sodium thiomethoxide led to a modest increase in its formation.

We expected to remove these byproducts by recrystallization, as had been done in the multiple-step process. Disconcertingly, however, the related substances cocrystallized tenaciously with pergolide, as free base or as a salt. The clear advantages of this new process would be largely lost if effective purification could not be achieved. Under most circumstances, chromatography on a production scale is undesirable from a cost and throughput basis. In the case of pergolide, however, the high potency that makes handling potentially hazardous also makes production volume relatively modest and of high value. Chromatography, then, is a more acceptable alternative.

Figure 5. The radial flow column used in the tandem chromatography purification. The column was needed to remove impurity 9 from the reaction mixture.




Purification by tandem chromatography

Using silica gel chromatography, we eventually separated all impurities except 9 (13). With additional investigation, we discovered that byproduct 9 was strongly adsorbed on HP20ss resin. Although this resin did not produce good separation between pergolide and its other byproducts, we took advantage of its adhesion and designed a tandem chromatographic process. Crude pergolide, as a dried solid or a damp filter cake (no dust), was dissolved in a mobile phase of 17% acetonitrile and pH 3 phosphate buffer and then eluted through a radial flow column (Figure 5); this type of column design allows high capacity and throughput. The eluant was pumped under pressure to the outside of the cylinder and through the resin to an inner collecting cylinder. The strongly adsorbed 9 was subsequently flushed from the resin by increasing the ratio of acetonitrile, regenerating the column for another cycle.

The acetonitrile from the eluate containing semipurified pergolide was removed by distillation. After the pH was adjusted to >9, the pergolide mixture was extracted into chloroform and eluted by HPLC. The production-size HPLC column was packed with high-performance silica gel using a mobile phase of chloroform, methanol, and methanesulfonic acid. Chloroform was not our preferred choice because it is toxic, but the chloroform/methanol/methanesulfonic acid eluant allowed substantially better loading capacity than the other halocarbons did (14). We also were able to efficiently recycle chloroform for reuse within this chromatographic system.

Crystallizing success

The eluate from the silica gel column was stirred in a dilute solution of aqueous ammonia. The resulting aqueous layer was separated, and the organic layer was concentrated by distillation. The concentrate was diluted with about 12 vol methanol, and 1 equiv methanesulfonic acid was added. The excess methanol was distilled to ~5 vol, 7 vol of methanol was added to a total of 12 vol, and the mixture was distilled back to 5 vol. This procedure was repeated twice with the addition and distillation of a similar volume of isopropanol. This multiple distillation-crystallization method from methanol/isopropanol was used to ensure the removal of residual chloroform and the isolation of the desired crystal form of pergolide mesylate. The dried product, an off-white solid of purity equal to or better than that produced via the multiple-step sequence, was obtained in 71% yield based on dihydrolysergol (15-17).

Thus, our basic objectives were achieved. Dihydrolysergol and pergolide mesylate are the only dry solids that must be handled, even with the necessary chromatography step. The yield more than tripled, and chemical processing time was reduced to two days.

This quaternization/demethylation method has been developed only for this process. Nevertheless, its excellent performance suggests that this chemistry may be of much wider scope than presented by the Hutchins and Dux (9).

References

  1. Bianchine, J. R. In Goodman and Gilman's The Pharmacological Basis of Therapeutics, 7th ed.; Gilman, A. G.; Goodman, L. S.; Rall, T. W.; Murad, F., Eds.; MacMillan: New York, 1985; pp. 473-90.
  2. Seeman, P.; Van Tol, H.H.M. Curr. Opinion Neurolog. Neurosurg. 1993, 6, 602-8.
  3. Physician's Desk Reference, 50th ed.; Medical Economics: Montvale, NJ, 1996; p. 575.
  4. Kornfeld, E. C.; Bach, N. J. U.S. Patent 4 166 182, 1979.
  5. Misner, J. W. U.S. Patent 4 782 152, 1988.
  6. Sprankle, D. J.; Jensen, E. C. In Analytical Profiles of Drug Substances; Brittain, H. G., Ed.; Academic: New York, 1992; vol. 21, pp. 375-413.
  7. Rehacek, Z.; Sajdl, P. Ergot Alkaloids: Chemistry, Biological Effects, Biotechnology; Elsevier: New York, 1990; vol. 12, p. 59.
  8. Rehacek, Z.; Sajdl, P. Ergot Alkaloids: Chemistry, Biological Effects, Biotechnology; Elsevier: New York, 1990; vol. 12, p. 28.
  9. Hutchins, R. O.; Dux, F. J. J. Org. Chem. 1973, 38, 1961-62.
  10. Shamma, M.; Deno, N. C.; Remar, J. F. Tetrahedron Lett. 1966, 1375-79.
  11. March, J. Advanced Organic Chemistry. Reactions, Mechanisms, and Structure, 4th ed.; Wiley: New York; 1992; p. 339.
  12. Aldrich Catalog Handbook of Fine Chemicals, 1994-1995; Aldrich: Milwaukee, WI, 1994; p. 589.
  13. Kennedy, J. H. Presented at the Midwest Pharmaceutical Process Chemistry Consortium, Indianapolis, IN, October 1993.
  14. Kennedy, J. H. In Organic Process Research and Development in press.
  15. Misner, J. W. U.S. Patent 5 463 060, 1995.
  16. Misner, J. W. Presented at the "Technical Achievements in Organic Chemistry" symposium at the 210th National Meeting of the American Chemical Society, Chicago, IL, August 1995; paper 118.
  17. Misner, J. W.; Kennedy, J. H.; Biggs, W. S. Organic Process Research and Development, in press.


ChemCenter Pubs Div. Home Page

Copyright © 1996 by the American Chemical Society.