The 14-hydroxyl substituted opioid antagonists, such as naltrexone (1) and naloxone (2), are valuable medications for the treatment of opiate abuse,¹ opiate overdose,² and alcohol addiction.³ However, the ever increasing demand for these medicinal opioids, coupled with the findings that delta subtype selective derivatives may be useful immunosupressants⁴ and also prevent the development of tolerance to morphine,⁵ has placed a premium on thebaine (3), the common starting material⁶ that has only a low natural abundance in opium.⁷

Obviously, the replacement of 3 as the starting material with a more available opiate, such as codeine (4) would alleviate the pressure on thebaine. As part of our program directed toward this goal, we recently reported a novel and practical synthesis of 3 from 4.⁸ Codeinone (5), which can be prepared through simple Oppenhauer oxidation of 4⁹, was shown to undergo enol ether formation to 3 by treatment with tBuOK, 18-crown-6, and Me₂SO₄, thus removing the need for the isolation of 3 from opium. Peracid oxidation of 3 gives the key intermediate 14-hydroxycodeinone (6),^10,11,12 from which the important antagonists can be prepared.⁶ We now wish to present our studies into the oxidation of 5 at the allylic position to give 14-hydroxycodeinone (6) directly, and thus remove the requirement for the preparation of thebaine as an intermediate in the syntheses of the antagonists.

The oxidation of 5 to 6 can be accomplished by a variety of 2-step procedures via formation of a diene system followed by oxidation.^13,14 Obviously, a direct procedure would offer the advantages of less synthetic steps and remove the necessity for the isolation of a diene intermediate. Previous attempts at such a direct oxidation have met with limited success due to competing oxidation at other positions on the opioid skeleton.^14,15 The only direct method resulting in a reasonable yield involves the use of MnO₂;¹⁶ treatment of codeine with MnO₂ leads to codeinone, however prolonged treatment gives 6 in 30% yield. Although this procedure was subsequently improved using a 3-silyl ether derivative (59%)¹⁷, the large quantities of MnO₂ employed limits its practicality for industrial scales. In an attempt to develop a practical method for this transformation, we investigated the treatment of 5 with a variety of readily available and easily handled oxidizing agents, with the ultimate aim of discovering a procedure that could be conveniently performed on commercial scales.

Peracid Mediated Oxidations

Initial studies concentrated on the fact that in acidic media, codeinone (5) exists in equilibrium with neopinone (7).¹⁸ It was envisaged that the different electronic character of the C-ring double bonds in 5 and 7 may allow selective oxidation of 7 with peracids, whereas the electron deficient double bond in 5 would not undergo oxidation. It was found that treatment of an equilibrium mixture of 5 and 7 under the performic acid conditions described by Iijima et al. for thebaine,¹⁰ led to low yields of 6 (Table 1, entry 1).

Although somewhat encouraging, the fact that the balance of the material could not be recovered suggested competing oxidations to give water soluble products. A range of peracids, with differing steric and electronic properties were therefore investigated in order to determine if the nature of the peracid affected product distribution and the results are summarized in Table 1.

As can be seen from the table, all reactions gave rise to only minor yields of 6, the greatest yield (37%) was obtained by the use of dimethylperacetic acid (entry 10). The low yields make it difficult to draw any definitive conclusions, however the fact that dimethylperacetic acid gave a higher yield than peracetic acid (entry 5) tends to suggest that steric bulk may be one factor in minimizing unwanted side reactions. This however, is not consistent with the poor results obtained with the even more bulky tri-substituted peracetic acids (entries 9 and 11). In all reactions the side products were water soluble and could not be isolated.²⁰ Although strong acid was included in all cases, the possibility of N-oxidation as one side reaction cannot be dismissed,²¹ and therefore the action of peracids on the corresponding carbamate protected derivative (8)⁸ was investigated (Table 2).

Surprisingly, treatment of 8 with peracids led to none of the desired 14-hydroxyl substituted product (10). This is however consistent with previous findings that basic and non-basic opioid derivatives can react very differently when treated under the same conditions.²² Little starting material could be recovered in any of the reactions and it was concluded that alternative oxidations (possibly Baeyer-Villiger²⁰) competed successfully.

Metallic Oxidants

The possibility that Baeyer-Villiger reactions of 7 and 9 were successfully competing with the desired oxidations above, suggested that the optimal reagent for this transformation will be one that is not capable of performing a Baeyer-Villiger reaction. In addition, it was concluded that some method of regiocontrol may be necessary to favor the desired oxidation over benzylic oxidation at C-10.²³ As it has been shown that the N-lone pair of opioids occupies a position above the C-ring,²⁴ it was considered that an oxidizing metallic species coordinated to the N-lone pair of 5 may be suitably positioned close to the desired site of oxidation thus favoring oxidation at this position. (Figure 1).

Encouraged by the recent report that a metallic species coordinated to a polyaniline ligand together with a cooxidant gave rise to allylic oxidation,²⁵ we decided to investigate the application of such a procedure to 5, where the opioid tertiary amine could act as the amino ligand. Thus 5 was treated with a range of readily available metallic oxidants, and the results are summarized in Table 3.

A complex mixture of materials resulted from all reactions which made isolation and identification of the products difficult. MS analysis of the mixtures showed signals corresponding to various oxidized products, however it was decided that further analysis was not useful in most of the cases. Of all the reactions, only the cobalt mediated oxidation (entry 2) appeared promising. Although giving just a minor yield of 6, it was noted by MS that the major side reaction appeared to be iodination when periodate was employed, and the loss of material with the use of Oxone was shown to be due to direct reaction of Oxone with 5.²⁸ In order to remove the complications obviously caused by the cooxidant, it was decided to investigate the reaction using stoichiometric quantities of Co(OAc)₃, a reagent that can be easily formed by the ozonolysis of an acetic acid solution of Co(AcO)₂·4H₂O.²⁹ The results of these Co(III) mediated oxidations are summarized in Table 4. It was found that treatment of 5 with 2 equivalents of Co(AcO)₃ in acetic acid at room temperature led to a 38% yield of the desired 6 in a very convenient reaction (entry 1). Further study of the reaction showed that the initial addition of 1 equivalent of Co(OAc)₃ and stirring for 24h, followed by the addition of a second equivalent and stirring for a further 4h led to an improved 51% yield of 6 (entry 2). Yields could not be improved further due to competing N-oxidation to N-formyl and N-nor derivatives (detected by MS), however the presence of the nitrogen lone pair was shown to be essential as the addition of 1 equivalent of H₂SO₄ inhibited the reaction (entry 4), and carbamate 8^30,8 was inert to the reaction conditions (entry 7). These results suggested that the basic tertiary amine was indeed acting as a ligand for the cobalt and directing oxidation to the correct position, however when coordinated, oxidation of this amino ligand was a competing reaction.³¹ Although allylic oxidation has been observed with Co(III) salts,^32,33 the reaction is less favorable than either benzylic³⁴ or aromatic oxidation³⁵ underscoring the importance of coordination of the metal to the teriary amine³⁶.

Interestingly, the treatment of 5 with Co(F₃CCO₂)₃³⁵ led only to a complex mixture from which no product could be readily identified, clearly showing the dramatic influence of the carboxylate ligands employed. The current work confirms that the direct oxidation of 5 to 6 cannot be performed with the vast majority of common oxidizing agents, however we have shown that this transformation can be conveniently performed by the use of Co(AcO)₃ in acetic acid at room temperature. This simple procedure gives comparable yields to the improved MnO₂ oxidation¹⁷ but, importantly, offers a truly practical method for the direct oxidation of 5 to 6.

In summary, we have developed a practical 1 step synthesis of 14-hydroxycodeinone from codeinone by the use of Co(OAc)₃ in 51% yield without the need for chromatography.