Scalable synthesis of a nucleoside phosphoramidate prodrug inhibitor of HCV NS5B RDRP: Challenges in the production of a diastereomeric mixture

Article abstract of DOI:10.1021/op5003837

A scalable process is described for the synthesis of 2′-C-methylguanosine-5′-[2-[(3-hydroxy-2,2-dimethyl-1-oxopropyl)thio]ethyl-N-benzylphosphoramidate], a nucleotide prodrug inhibitor of hepatitis C virus NS5B polymerase. The route features the use of ph

Full text of DOI:10.1021/op5003837

Article
pubs.acs.org/OPRD
Scalable Synthesis of a Nucleoside Phosphoramidate Prodrug
Inhibitor of HCV NS5B RdRp: Challenges in the Production of a
Diastereomeric Mixture
Benjamin A. Mayes,* Jingyang Wang, Jeevanandam Arumugasamy, Kannan Arunachalam, Erkan Baloglu,
David Bauer, Alan Becker, Narayan Chaudhuri, Roberta Glynn, G. Mark Latham, Jie Li, Jinsoo Lim,
Jia Liu, Steve Mathieu, F. Patrick McGarry, Elodie Rosinovsky, Adrien F. Soret, Alistair Stewart,
and Adel Moussa
Idenix Pharmaceuticals Inc., 320 Bent Street, Cambridge, Massachusetts 02141, United States
S
* Supporting Information
ABSTRACT: A scalable process is described for the synthesis of 2′-C-methylguanosine-5′-[2-[(3-hydroxy-2,2-dimethyl-1-
oxopropyl)thio]ethyl-N-benzylphosphoramidate], a nucleotide prodrug inhibitor of hepatitis C virus NS5B polymerase. The
route features the use of phenylboronic acid to transiently protect the 2′,3′-hydroxyls of 2′-C-methylguanosine under mild
conditions. The requirement to produce a 1:1 phosphorus diastereomeric mixture precluded the use of traditional crystallization
techniques. High sensitivity of the drug substance to acidic and basic conditions and its preferred solubility in mixed aqueous−
organic solvents presented additional processing challenges. The use of reverse phase chromatography for the final purification
was eliminated by the development of a dual liquid−liquid extraction protocol, which removed both non-polar and polar
impurities whilst maintaining the 1:1 diastereomeric ratio. Ethylene sulfide, a potential genotoxic impurity that was observed at
significant levels in the original procedure, was controlled to <10 ppm in the final product. This process produced 20 kg of drug
substance in 59% overall yield, with >99% purity in the requisite 1:1 diastereomeric ratio.
perspective, the API typically coagulated and became sticky
prior to full solubilization on dissolution from dry powder.
Although the in vitro HCV replicon assay activities of the
separate Rp and Sp diastereomers were similar (within 3-fold),
the pure separated isomers did indeed display marked physical
differences: for example, in contrast to the Sp isomer, the
enriched Rp isomer was reluctant to crystallize in a range of
solvent systems and was only ever isolated as an amorphous
solid by evaporation to dryness; however, the relatively well
behaved Sp isomer was amenable to the growth of single
crystals suitable for structural determination by X-ray analysis
(Figure 2).
In addition to these physical characteristics, the fragile nature
of the prodrug moiety imparted problematic chemical
characteristics, in particular, sensitivity to both acid and base.
Even trace levels of acid or base were determined to be
detrimental in the presence of a protic solvent, typically
resulting in the formation of the free 2′-C-methylguanosine
nucleoside 3, the nucleoside 5′-monophosphate, the simple
benzylphosphoramidate of 2′-C-methylguanosine, or a nucleo-
side phosphorus triester via solvolysis of one or both of the S-
acyl-thioethyl or benzylphosphoramidate moieties.
INTRODUCTION
■
2′-C-Methylguanosine-5′-[2-[(3-hydroxy-2,2-dimethyl-1-
oxopropyl)thio]ethyl-N-benzylphosphoramidate] (IDX184, 1)
is a nucleoside phosphoramidate prodrug inhibitor of hepatitis
C virus (HCV) which has reached Phase 2 clinical trials (Figure
1).1 Post oral administration, metabolism in the liver to the
nucleoside monophosphate, followed by sequential phosphor-
ylation by cellular kinases, produces 2′-C-methylguanosine
triphosphate (2′-MeG-TP, 2), which is a potent inhibitor of
HCV NS5B RNA-dependent RNA polymerase (RdRp) in vitro
(1b WT IC50 = 0.292 μM).2 The etiology of hepatitis C
infection and the development of direct-acting antiviral
treatments, in particular nucleotide prodrug inhibitors of
HCV RdRp, are thoroughly documented.3 As an advanced
clinical candidate, a practical synthetic route was required to
enable production of IDX184 with high purity on a multi-
kilogram scale.
An atypical active pharmaceutical ingredient (API), 1
presented multiple synthetic and processing challenges. Fore-
most, as an S-acyl-thioethyl benzylphosphoramidate prodrug of
2′-C-methylguanosine monophosphate, 1 is a mixture of
phosphorus diastereomers in a 1:1 ratio. In addition to the
anticipated differences in physical properties between the
separate Rp and Sp isomers, 1 displayed a somewhat
amphiphilic nature whereby the polar nucleoside was counter-
balanced by the relatively more lipophilic prodrug. As a result,
the solubility of 1 in either pure aqueous or organic systems
was generally low, and aqueous/organic mixtures were required
to achieve appreciable solubilites.4 Notably, from a handling
Additionally, chemical degradation of 1 was accompanied by
the generation of ethylene sulfide (ES), which under stressed
1
conditions was observable even by H NMR (103 ppm). ES is
classified as a potential genotoxic impurity (PGI), and as such,
its formation as a process impurity or degradant was a key issue
Received: December 10, 2014
© XXXX American Chemical Society
A
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 1. IDX184 (1) and 2′-MeG-TP (2).
Figure 2. ORTEP drawing of the X-ray crystal structure of Sp diastereomer, showing crystallographic numbering system. Anisotropic atomic
displacement ellipsoids for the non-hydrogen atoms are shown at the 50% probability level. Hydrogen atoms are displayed with an arbitrarily small
radius.
a
Scheme 1. Discovery Chemistry Synthesis of Prodrug 1
a
Conditions: (i) a) Me3CCOCl, pyridine, −15 °C; b) SiO2 chromatography (32%); (ii) a) BnNH2, CCI4, 5 °C; b) SiO2 chromatography (99%);
(iii) a) TFA, DCM, 25 °C; b) SiO2 chromatography; c) C-18 chromatography; d) lyophilization (39%).
in the isolation of final API suitable for use in prolonged clinical
trials.
This article describes the development of the synthetic
process utilized to produce 1 on a 20 kg scale. Various
B
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
a
Scheme 2. Synthesis of Analogous 2′-C-Methylcytidine Prodrug 10 Using PBA Transient Protection
a
Conditions: (i) a) PhB(OH)2, pyridine, reflux; b) H-phosphonate 4, EDCI·HCl, MeCN; c) BnNH2, CCl4; d) 20% w/w citric acid (aq) (53%); (ii)
AcCl, EtOH (61%).
a
Scheme 3. Synthesis of Prodrug 1 on Small Scale Using PBA Transient Protection
a
Conditions: (i) a) PhB(OH)2, MeCN, Na2SO4 (s, 3.5 equiv), reflux; b) H-phosphonate 4, Me3CCOCl, pyridine; c) BnNH2, CCl4; d) 22% w/v
citric acid (aq) (66%); (ii) a) AcCl (2 equiv), EtOH, 60 °C; b) NaHCO3(s) neutralization; c) MTBE trituration; d) C-18 chromatography (65%).
challenges encountered with this unusual API are highlighted,
along with the methods which were implemented for their
circumvention.
reported on a small scale (<1 g). The original procedure used
in discovery allowed isolation of the first milligram quantities of
1 in three steps from 2′-C-methylguanosine 3 (Scheme 1).5
The unprotected nucleoside was initially coupled with H-
phosphonate salt intermediate 4 in low yield (32%). Although
transformation of H-phosphonate 5 to phosphoramidate 6 was
relatively efficient, subsequent terminal trityl group depro-
tection with TFA was not, and the final prodrug 1 was obtained
RESULTS AND DISCUSSION
■
Synthetic Approaches. Two syntheses of phosphor-
amidate prodrug 1 or its separate diastereomers have been
C
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
in a mere 12% overall yield from 3. Silica gel chromatography
was employed at all steps, and additionally, the target
phosphoramidate 1 underwent chromatography on reverse
phase silica gel to achieve acceptable purity, followed by
lyophilization.
via an Atherton−Todd reaction.9 This switch in order of
addition relative to the original route was determined to be
highly beneficial, as charging the benzylamine second invariably
led to significant levels (∼10%) of the undesired P-OH side
product prior to installation of the phosphoramidate.
Performed in this manner, the Atherton−Todd reaction
efficiently and reproducibly provided a 1:1 mixture of
diastereomers at phosphorus by in-process LC analysis.
On addition of aqueous citric acid and MTBE, the PBA
protection was completely and rapidly hydrolyzed in situ,
revealing the phosphoramidate 6. Sequential mild aqueous
washes at pH 4 and pH 8 allowed removal of pyridine, pivalic
acid, benzylamine, and PBA-related byproducts whilst main-
taining product integrity. Isolation of 6 from the resultant
organic layer was then achieved on addition of further MTBE,
which induced precipitation of both product diastereomers. It
was typically observed that the Rp diastereomer was more
reluctant to precipitate than the Sp; however, on prolonged
stirring, the phosphorus isomer ratio invariably met the
required 45:55 specification.10 Filtration of the precipitated
solids and washing with the next process solvent provided 6
with >95% LCAP11 purity with a 1:1 diastereomer ratio in 66%
yield from nucleoside 3. As with 2′-C-methylcytidine 7, this
PBA transient protection methodology resulted in a 2-fold
increase in coupling yield relative to the analogous coupling
with the 2′,3′-hydroxyls unprotected.
It was immediately evident from the discovery chemistry
procedures that removal of the terminal trityl protecting group
using TFA/DCM was impractical due to a high level of product
degradation. Similar to the process with the analogous 2′-C-
methylcytidine prodrug 9, dry HCl was instead found to be a
superior alternative. Accordingly, treatment of 6 with 2 equiv of
dry HCl, generated via reaction of acetyl chloride with ethanol,
promoted complete deprotection of the trityl group at 60 °C in
0.5−1 h. Neutralization to pH 5−6 was performed on addition
of solid NaHCO3, thus avoiding aqueous systems, and after
filtration through Celite, the filtrate was evaporated to a solid
prior to trityl byproduct removal on trituration with MTBE.
Purification of the crude solid 1 (typically 80−85% LCAP)
was then undertaken using chromatography on C-18 silica gel
(11:1 w/w silica:crude ratio). The crude material was dissolved
in a mixture of THF, water, and saturated aqueous NaHCO3, to
ensure the impure 1 was loaded onto the column at a relatively
neutral pH.12 Elution was performed stepwise with seven
increasingly non-polar solvent systems from 1.5 to 25%
MeCN/water (∼300 vol total).13 A large volume of high
aqueous content systems was required to remove multiple
close-running impurities. Both diastereomers eluted in the same
system (25% MeCN/water); however, the Rp diastereomer
eluted somewhat earlier than the Sp, and accordingly, judicious
fraction selection by LC analysis was required to achieve high
chemical purity whilst maintaining the desired 1:1 phosphorus
isomer ratio. After evaporation of the appropriate MeCN/water
fractions, the target prodrug 1 was obtained in 65% yield with
>98% LCAP purity in 1:1 diastereomer ratio.
The second synthesis described access to the individual
diastereomers of 1, in which a racemic benzyl-pentafluor-
ophenyl-phosphoramidate intermediate was formed prior to
phosphorus enantiomer separation and subsequent coupling
t
with 6-O-methyl-2′-C-methylguanosine using BuMgCl-medi-
ated 5′-hydroxyl activation.6 In this case, coupling yields with
the two separated enantiomeric intermediates were a disparate
59 and 36%, respectively, which potentially indicated that
formation of 1 in a 1:1 ratio at phosphorus may be problematic
via this approach. Additionally, the formation of the key
racemic pentafluorophenyl intermediate in low yield (38%) and
the use of 6-O-methyl-protected 3, which necessitated an extra
methoxypurine-to-guanine conversion, made this methodology
unattractive to pursue.
Preceding investigations into this 2′-C-methylguanosine
prodrug, an analogous nucleotide prodrug inhibitor of HCV
NS5B RdRp, 10, was under development, the synthesis of
which employed the use of phenylboronic acid (PBA) as an
effective transient means to protect the 2′,3′-hydroxyls of 2′-C-
methylcytidine 7 (Scheme 2).7 This approach was determined
to be superior to the use of either unprotected or acetonide-
protected 2′-C-methylcytidine. Although this PBA method-
ology was only demonstrated on a 100 g scale, its application to
the synthesis of 1 was appealing: only two synthetic steps were
required from two advanced building blocks, 3 and 4, both of
which were readily accessible on a multi-kilogram scale.8
PBA Route. The procedure developed to produce
phosphoramidate 1 is shown in Scheme 3. Protection of
nucleoside 3 with PBA was performed in refluxing MeCN in
the presence of anhydrous sodium sulfate to sequester the 2
mol of water generated during the condensation. Whereas with
2′-C-methylcytidine 7 either MeCN or pyridine was utilized as
1
solvent to achieve high conversions (>97% by H NMR), in
this case, MeCN was preferred in order to avoid the use and
subsequent evaporation of large volumes of pyridine on scale.
Regioselectivity of the boronate formation was confirmed by
2D NMR, and the hydrolytic behavior of 2′-C-methylguano-
sine-2′,3′-cyclic boronate was determined to be similar to that
of the analogous cytidine, indicating its suitability for use as a
transient protecting group. In order to avoid premature
hydrolysis of the cyclic boronate 11, an additional charge of
Na2SO4 was performed prior to the addition of H-phosphonate
salt 4 (1.5 equiv) dissolved in MeCN. Coupling was then
effected via a mixed phosphorus anhydride intermediate on
addition of a solution of pivaloyl chloride (5 equiv) in pyridine
(10.5 equiv) to a chilled mixture of boronate-protected
nucleoside and 4, generating the H-phosphonate 12. Coupling
using pivaloyl chloride/pyridine at <15 °C, or alternatively the
carbodiimide reagent EDCI at 40 °C, required in both cases
similar stoichiometries (5 equiv) for good conversion: in
contrast to the cytidine analogue, in this case the more cost-
effective and simpler to handle acid chloride was preferred for
use on scale.
The overall yield of 1 from nucleoside 3 of 40−45%
compared favorably with the 12% yield via the original
unprotected approach, and this PBA methodology was utilized
to produce the first kilogram batches of 1. Despite this
improvement, multiple challenges were observed which needed
to be addressed to enable production of larger quantities:
Whereas in the original discovery synthesis the unprotected
H-phosphonate intermediate 5 was isolated, here the boronate-
protected H-phosphonate 12 was used in situ: the addition of
benzylamine followed by carbon tetrachloride successfully
resulted in rapid conversion to the benzylphosphoramidate
D
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 3. Major impurities observed in deprotection of 6.
(1) It was imperative that the initial boronate protection
proceed to completion. In one instance where the
agitator speed was too high, a significant portion of the
nucleoside 3/MeCN suspension was determined to have
coated the walls of the reactor prior to reaction with
PBA, leading to a non-representative in-process
protection sample. The subsequent coupling step
consequently performed poorly (∼30% LCAP), which
negatively impacted both the quantity and quality of 6.
(2) Although it is utilized only as a reagent (12 equiv), the
use of the class 1 solvent CCl4 in the Atherton−Todd
reaction was not ideal from regulatory or environmental
perspectives.
(3) The AcCl/EtOH deprotection was vastly superior to the
original TFA protocol; however, it suffered from the
concomitant generation of P-OEt impurity 13 at up to
10% LCAP, dependent upon the amount of acid,
temperature, and reaction time. This impurity was not
trivial to remove by chromatography, due to its polarity
being similar to that of 1, which partially dictated the
methodical approach to the elution systems.
aqueous/organic product solutions at elevated temper-
atures. Even trace levels of acid, from incomplete
NaHCO3 neutralization, could induce product degrada-
tion on prolonged distillation of the elution solvents. The
RP chromatography itself was not efficient nor
reproducible, as based on the quality of the input crude
material, product fraction selection was dictated by both
chemical and diastereomeric purity to achieve a 1:1
isomer ratio. Indeed, in one instance, RP chromatog-
raphy was completely ineffective, as the product was not
retained on the column, eluting in the first 1.5% MeCN/
water system. It was determined that the use of heptane
in place of MTBE in the precipitation of 6, in order to
achieve a more precise 1:1 diastereomer ratio, had also
precipitated an unidentified non-UV-active material
which negatively impacted the performance of the final
purification. Additionally, RP silica gel brand variability
and accessibility on scale were further hindering factors,
as was the fact that 50% of the estimated cost of goods
for 1 was due to the RP silica gel itself.
(7) The existing chromatographic purification and subse-
quent evaporation to dryness of product fractions
resulted in the observation of ES at significant levels
(range 400−3000 ppm) in the final API 1.14 Because ES
is a PGI, these levels prohibited the utilization of several
batches of 1 for clinical trial durations longer than 1
month at the proposed 200 mg daily dose.15 A protocol
was therefore required to reduce the ES level whilst
maintaining the overall chemical purity and 1:1
diastereomeric ratio. Based on the observation that ES
levels did not increase over time on storage,16 it was
determined that, if 1 were produced with acceptable
levels of ES at t = 0 h, then prolonged shelf life would not
present an issue.
(4) Neutralization of the in situ generated HCl on
completion of the deprotection reaction by solid
NaHCO3 was far from efficient, and the subsequent
filtration through Celite on scale proved to be slow.
(5) Processing of the ethanolic filtrate by concentration to
dryness and trituration of the resulting sticky solid to
remove the trityl byproduct was not feasible for larger
production equipment.
(6) The use of column chromatography as a final purification
method presented multiple practical hurdles. Because it
was determined that normal phase silica gel was not
suitable to provide effective separation of impurities,
chromatography on reverse phase (RP) silica gel was
utilized, which required evaporation of large volumes of
E
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Initial ES analysis indicated a substantial reduction (158 ppm)
and subsequent further EtOH/water coevaporation actually
surpassed target ES levels (30 ppm). An improvement in LCAP
purity to 98.3% occurred, and importantly, the 1:1 phosphorus
diastereomer ratio had been unaffected.
(8) The final physical form of 1 obtained on evaporation to
dryness presented some degree of diastereomeric
inhomogeneity, presumably due to the different precip-
itation rates of the two isomers and their non-uniform
distribution within the resultant amorphous API.
Encouraged by the 2-fold benefit of this extractive
purification, namely a >95% reduction in ES and removal of
polar impurities into the aqueous layer, this protocol was
pursued. In the short term it was determined to be
advantageous as an effective means to extract 1 from water/
MeCN product fractions post RP chromatography. On addition
of EtOAc, the organic phase containing EtOAc and MeCN
(and water) was separated, thus reducing the water distillation
volume by 80−90%. Based on the fact that 1 can remain
dissolved in MeCN/water 1:1 mixtures at 10% w/v, it was
envisioned that selection of a suitable low-polarity organic
solvent may enable the inverse product partition, forcing 1 into
the “aqueous” layer. Toluene or EtOAc/heptane were
determined to be feasible, allowing extraction of multiple
low-polarity impurities into the respective organic phases.
Toluene was adjudged to be more effective, and on a small
scale, a mixture of toluene, MeCN, and water in 2:3:2
volumetric ratio was demonstrated on neutralized crude 1 to
produce a suitable phase separation, removing the low-polarity
impurities 15, 16, and 17 and unreacted 6 whilst maintaining a
1:1 diastereomer ratio in the resultant aqueous layer. This
second extractive purification eliminated the existing problem-
atic MBTE trituration to remove the trityl group byproduct 16,
which had resulted in the formation of sticky solids with
difficult handling properties.
The P-OEt side product 13 was the remaining impurity of
concern, as it was not efficiently purged by either the EtOAc/
MeCN/water or the toluene/MeCN/water extractions, due to
its polarity being similar to that of product 1. Formed as a side
product by acidic ethanol solvolysis of 1, modification of the
trityl deprotection reaction to be conducted under milder
conditions was deemed to be the most feasible method to
achieve the desired specification of <0.50% LCAP for the sum
of both diastereomers.21 Accordingly, the amount of dry HCl
generated on addition of AcCl to EtOH was reduced from 2 to
1.5 equiv, and the amount of EtOH was substantially reduced
by utilizing MeCN as a cosolvent in 1:1 MeCN/EtOH
volumetric ratio. Lowering the reaction temperature from 60
to 40 °C substantially decreased the deprotection rate,
accommodating longer processing times on increased scale.
In place of solid NaHCO3, the reaction mixture was treated
with Dowex 66 basic resin, thus avoiding the slow filtration
through Celite and reducing the potential for incomplete
neutralization.
Accordingly, multiple modifications of the existing PBA
methodology were required for scale up to fixed equipment.
Scale Up Challenges. Purification and Isolation of 1. It
was apparent that a new deprotection, workup, and isolation
protocol was required to remove various process impurities
whilst avoiding the use of RP chromatographic purification,
evaporation to dryness procedures, and the resultant formation
of sticky non-homogeneous solids. Typical impurities observed
after the standard trityl deprotection reaction conditions
included various product-related species of polarity higher
and lower than that of 1, along with the three-membered ring
ethylene sulfide 18, which was the impurity of most immediate
concern (Figure 3). International Agency for Reseach on
Cancer (IARC) monographs on the evaluation of carcinogenic
risks to humans categorize ES into Group 3, “the agent is not
classifiable as to its carcinogenicity to humans”; however, for
the purposes of this program, ES was treated as a confirmed
genotoxic impurity (GTI).17 The existing FDA18 and EMA19
guidances on approaches to deal with GTIs have recently been
augmented by ICH guideline M7,20 in which recommendations
for assessment of actual and potential mutagenic impurities in
final drug substances are outlined. Under this framework, ES
was assigned to Class 3, categorized as an impurity which
exhibited an alerting structure, unrelated to the drug substance,
without mutagenicity data. In this case, a control strategy was
pursued in order to produce 1 with ES at levels below the
generic threshold of toxicological concern (TTC). The
anticipated clinical trial duration for IDX184, as with related
direct-acting antiviral HCV therapeutics, was 1−3 months,
which dictated a maximum acceptable ES intake of 20 μg per
day. At the intended 200 mg daily dose, this corresponded to
an acceptable ES level of up to 100 ppm in the final drug
substance 1.
Initially, various potential methods to reduce ES content in 1
were investigated. Despite the volatility of ES (bp 56 °C),
extended vacuum drying of the amorphous powder at 40 °C
did not achieve target reductions. Silica-supported amine and
thiol nucleophlic scavengers were unsatisfactory and, in the case
of the amine functionality, actually promoted degradation of 1.
Conversely, either Ar or N2 degassing of aqueous/organic
product solutions at 40 °C, followed by repeated co-
evaporation of 1 with EtOH/water 1:1 mixtures or spray-
drying under mild conditions (outlet temperature 60−100 °C),
resulted in acceptable ES levels (<100 ppm) on gram scale.
Additionally, a liquid−liquid extraction approach was ex-
plored whereby the poor solubility of 1 in aqueous or organic
solvents was modulated on introduction of a cosolvent, which
was selected to effect partition of 1 into the predominantly
organic layer, whilst maintaining phase separation from the
predominantly aqueous system. Accordingly, a mixture of
EtOAc, MeCN and water in 6:5:3 volumetric ratio exhibited
suitable properties and phase separation was observed to be
rapid when conducted between 25 and 40 °C. Extraction of
impure 1 (92% LCAP, 3000 ppm ES) was demonstrated on
100 g scale and after evaporation of the “organic” phase, which
contained ∼20% of the input water, followed by coevaporation
with EtOH/water, 1 was obtained with >90% mass recovery.
Altogether, these modifications to the deprotection reaction
and workup, along with implementation of dual liquid−liquid
extractive purifications, allowed isolation of 1 without RP
chromatography on 50 g scale with 99.6% LCAP purity, 0.19%
LCAP 13, in 1:1 diastereomeric ratio. Subsequent trituration of
the agglomerated solids in IPAc at 40 °C resulted in a
significantly more homogeneous, easily filtered amorphous
powder, which maintained the same 1:1 diastereomer ratio.
Final analysis showed an ES level of 12 ppm; thus, this
procedure formed the basis of a process for scale up to fixed
equipment.
Production of 1 in Fixed Equipment. The initial boronate
protection was performed on 8 kg of nucleoside 3, with PBA
and solid Na2SO4, suspended in refluxing MeCN. After 4.5 h,
F
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
1H NMR in-process analysis indicated 99% conversion to the
desired cyclic boronate 11. The second charge of solid Na2SO4
was omitted on this scale, as boronate hydrolysis was
determined to be negligible on proceeding directly to the
subsequent reaction with H-phosphonate 4.
Coupling using pivaloyl chloride/pyridine was performed at
8−17 °C as for the smaller scale runs, and after 2.5 h, analysis
indicated the LCAP ratio of P-H intermediate to 3 was 7.65:1,
at which point the reaction was considered complete.22 On
addition of benzylamine, the batch thickened but remained
stirrable, and after activation with CCl4, the conversion of P−H
intermediate 12 to the boronate-protected benzylphosphor-
amidate completed with >99% conversion.
Cleavage of the cyclic boronate with aqueous citric acid and
dilution with MTBE yielded three phases, a lower aqueous
phase, a middle phase containing solids, and an upper organic
phase, from which the lower aqueous was separated. After being
washed with aqueous NaHCO3/NaCl, the organics were twice
diluted with MTBE and partially distilled to remove residual
water (and MeCN), which was determined to negatively impact
the performance of the subsequent filtration. Addition of
further MTBE induced precipitation of both phosphorus
diastereomers: after 7 h, the Rp:Sp ratio was within specification
at 0.459:0.540; however, on stirring for a further 25 h, the ratio
of Rp:Sp was a more satisfactory 0.476:0.524. Filtration of the
solids, washing with MTBE and EtOH, and vacuum drying at
45 °C produced 20.7 kg of trityl-protected 6 with 97.7% LCAP,
0.41 wt % water, 0.19 wt % MBTE and 1.4 wt % EtOH, in 85%
purity adjusted yield.
system. A 2.8:1.0:1.3 overall volumetric ratio of EtOAc/
MeCN/water produced a clearly defined separation of phases,
and an organic product layer, containing EtOAc and MeCN
with an estimated 5% of the total water charged, was obtained
with 98.9% LCAP and 0.466:0.534 Rp:Sp diastereomer ratio.
The P-OEt 13 impurity was partially retained in the aqueous
layer, reducing the total level to 0.38% LCAP. A second
extraction of product from the aqueous phase was performed
using a 2.8:1.0:1.1 overall volumetric ratio of EtOAc/MeCN/
water, and a second organic product layer, which contained
EtOAc and MeCN with approximately 10% of the total water,
was separated. This product layer was significantly more dilute
than the first and contained 0.70% LCAP of P-OEt 13, 98.5%
LCAP purity of 1 with 0.425:0.548 Rp:Sp diastereomer ratio.
The two mixed organic product phases were combined, and
analysis of the residual aqueous solutions indicated 4.9% of the
theoretical product 1 yield remained unextracted.
A second deprotection reaction performed on the same scale
of input 6 (21.7 kg) proceeded in an identical manner, and the
two sets of organic extracts were combined to be further
processed as one batch (98.7% LCAP average). At this point,
prior reactions on smaller scale had involved distillation of the
product solutions to dryness before co-evaporating with EtOH/
water to ensure low ES content and triturating with IPAc. As
these operations were not practical on larger scale, a solvent
exchange of the EtOAc/MeCN25 extracts to isopropanol was
performed, which ensured complete removal of residual solvent
and water, prior to a second solvent exchange into IPAc for the
final trituration.
Accordingly, approximately 95% of the combined organic
product solution was evaporated, and five sequential
isopropanol charges and equivalent volume distillations were
performed. By the final isopropanol charge, a thick product
slurry was observed. Seven sequential IPAc charges and
equivalent volume distillations were then performed to
maximize recovery of 1 and maintain the desired 1:1
diastereomer ratio. Trituration of the resulting thick product
slurry in IPAc at 35 °C, then slow cooling to 0 °C followed by
filtration and tray-drying, allowed isolation of 1 with some
agglomerated but non-sticky particulates. Subsequent pin-
milling provided a uniform homogeneous amorphous powder
with the desired particle size distribution (D90 = 200−300 μm).
Residual solvent limits were determined to be <5000 ppm for
MBTE, isopropanol, and IPAc,26 and a <9 ppm value for ES
was achieved. The final Rp:Sp diasteremeric ratio was
0.457:0.543, which was effectively unchanged throughout the
purification processing. The constituent P-OEt 13 diaster-
eomers represented the largest single impurities and were
integrated at 0.26% and 0.13% LCAP, respectively. Accounting
for impurities ≥0.05%, the final purity of 1 was determined to
be 99.4% LCAP. The deprotection yield was improved on scale
to 69%, purity adjusted. Overall, starting from two 8 kg batches
of nucleoside 3, 20.2 kg of 1 was obtained, representing a 59%
two-step yield.
Deprotection of 6 was performed on 20 kg scale using the
new reaction conditions developed to minimize the level of P-
OEt 13 side product, with the modification that the reaction
temperature was further reduced as a conservative measure,
from 40 to 25 °C. The LCAP ratio of 1 to 6 was 2.64:1 at 5 h,
6.55:1 at 8 h, and 11.3:1 at 11 h, and the P-OEt 13 LCAP
increased from 0.24% to 0.38% to 0.52% over the same time
points.23 The reaction was quenched at 11 h to avoid further
formation of this impurity. Two charges of Dowex 66 resin
neutralized the mixture to target pH 6, and after filtration to
recover the resin, the product filtrate purity was 88.1% LCAP.
Extractions with the toluene/MeCN/water system to remove
impurities with RRT > 1, along with residual starting material 6,
were performed first, after distillation of 90% of the filtrate
solvents. A 1.0:1.9:1.4 overall volumetric ratio of toluene/
MeCN/water was utilized, and after separation of the lower
aqueous phase, which contained ∼20% of the total MeCN
charged, analysis indicated 97.7% LCAP purity with
0.465:0.535 Rp:Sp diastereomer ratio. A second extraction of
lower polarity impurities from the aqueous layer was performed
with a 1.0:1.4:2.4 overall volumetric ratio of toluene/MeCN/
water. This second aqueous phase, which contained ∼25% of
the total MeCN charged, was determined to have a slightly
improved purity (97.9% LCAP). The Rp:Sp ratio after the
second toluene extraction was unchanged at 0.465:0.535, as was
the total amount of P-OEt impurity 13 at 0.55% LCAP, with
contributions of 0.35% and 0.20% LCAP from its respective
diastereomers. Analysis of the toluene/MeCN extracts
indicated that the total amount of 1 contained in the first
and second organic phases was 6.5% of the theoretical product
yield.24
The use of CCl4 as a reagent (12 equiv) in the Atherton−
Todd transformation of P-H intermediate to the benzyl-
phosphoramidate 6, although efficient, was undesirable.27
During development and production, analysis of multiple
batches of 1 indicated that the required concentration limit (<4
ppm) was consistently achieved. With the intention of
switching CCl4 for an alternative chlorinating agent in the
Atherton−Todd reaction for the production of subsequent
clinical batches, small feasibility scale assessments were
The MeCN/water product solution was reduced in volume
by half prior to removal of impurities with RRT < 1 on
extraction of the product 1 with the EtOAc/MeCN/water
G
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
a
Scheme 4. Optimized Process for the Production of Prodrug 1 on a Multi-kilogram Scale
a
Conditions: (i) a) PhB(OH)2, MeCN, Na2SO4 (s, 2.5 equiv), reflux; b) H-phosphonate 4, Me3CCOCl, pyridine, 8−17 °C; c) BnNH2, CCl4, 5 to
25 °C; d) 21.3% w/w citric acid (aq) (85%); (ii) a) AcCl (1.5 equiv), EtOH: MeCN 1:1, 25 °C; b) Dowex 66 resin neutralization; c) toluene/
MeCN/H2O extraction; d) EtOAc/MeCN/H2O extraction; e) IPA solvent exchange; f) IPAc solvent exchange; g) IPAc trituration (69%).
undertaken. N-Chlorosuccinimide and 1,3-dichloro-5,5-
dimethylhydantoin (DCDMH) were determined to be suitable
alternatives, and proof of concept was successfully achieved on
5 g scale using 0.6 equiv of DCDMH. Importantly, the reaction
workup behaved similarly to that using CCl4, resulting in trityl-
protected 6 with a 1:1 diastereomer ratio.
The requirement to produce 1 as a 1:1 mixture of Rp:Sp
diastereomers precluded the use of typical crystallization
techniques. The original use of impractical, non-reproducible,
and expensive final purification via RP chromatography was
obviated. In its place, a dual extraction protocol was developed
to remove first non-polar and then polar impurities, and
judicious use of solvent exchanges avoided the formation of
sticky solids, whilst maintaining the 1:1 diastereomer ratio.
Additionally, the GTI ethylene sulfide was controlled to
levels <10 ppm, which allowed progression of the program into
clinical trials of the desired 1−3 month duration. Overall, 20 kg
of 1 was produced in one campaign from two 8 kg nucleoside
batches with 99.4% LCAP purity, in 0.46:0.54 Rp:Sp ratio, with
59% two-step yield. The associated PMI was reduced by 90%,
to 562 kg/kg, relative to the prior unscalable methodology.
It remains to be determined whether arylboronic acids and/
or esters in general represent a genotoxic hazard.28 Various
methods to control residual arylboronic acids in APIs have been
proposed,29 and the use of the N-methylglucamine-containing
polystyrene resin Amberlite IRA74330 was particularly effective
at removing PBA in related chemical processes in-house. In the
case of PBA itself and the cyclic boronate 11, no specific Ames
tests were conducted for these compounds; however, the boron
content of the final API 1 produced on 20 kg scale was
measured by ICP-OES to be 6 ppm, indicating a low risk of
residual PBA being present at meaningful concentrations.31
It is illuminating to calculate the overall process mass
intensity (PMI) metric for the production of 1, where PMI is
defined as the total mass of all reactants, reagents, and solvents
input during both reaction and purification per kg of product
output.32 Initial campaigns starting with 500 g−1 kg of
nucleoside 3 using the PBA transient protection, subsequent
deprotection of 6, and purification of 1 using RP silica gel
chromatography (Scheme 3) resulted in a cumulative PMI of
4831 kg/kg. The corresponding optimized process, developed
on 8 kg batch sizes of 3, which utilized extractive purifications
to avoid the use of RP silica (Scheme 4), achieved a cumulative
PMI of 562 kg/kg. While still relatively modest in absolutely
terms, these modifications represented a reduction in PMI of
almost 90%.
EXPERIMENTAL SECTION
■
General. 2′-C-Methylguanosine was obtained commercially.
HPLC was performed on a Waters 2965 Alliance separation
module equipped with a 2996 PDA detector, an Agilent 1100
series instrument equipped with a UV detector or equivalent
LC instrument. Method for 6: Zorbax Eclipse XDB C8 reverse
phase column (4.6 mm × 75 mm, 3.5 μm) at 28 °C; flow rate
1.4 mL/min; injection volume 8 μL; mobile phase A = 0.01 M
ammonium acetate aqueous pH 4.4 buffer, B = MeCN; run
time 20 min; gradient 5% to 80% B over 5.5 min, hold at 80% B
for 4.5 min; gradient 80% to 5% B over 3 min, hold at 5% B for
7 min; monitoring 254 nm. Method for 1: Waters XBridge C18
reverse phase column (4.6 mm × 100 mm, 3.5 μm) at 30 °C;
flow rate 1.0 mL/min; injection volume 10 μL; mobile phase A
= 95:5 H2O/MeCN, B = 95:5 MeOH/MeCN; run time 65
min; gradient 5% to 30% B over 10 min, hold at 30% B for 20
min; gradient 30% to 100% B over 25 min, hold at 100% B for
2 min, gradient 100% to 5% B over 8 min; monitoring 255 nm.
2′-C-Methylguanosine-2′,3′-O-phenylboronate 11. 2′-
C-Methylguanosine 3 (1.0 g, 3.36 mmol, 1.00 equiv) was
suspended in anhydrous MeCN (10 mL) between 20 and 25
°C under argon. PBA (0.43 g, 3.53 mmol, 1.05 equiv) was
charged in one portion, and the mixture was heated at reflux
(82 °C) for 3 h. The suspension was then cooled to 20−25 °C
CONCLUSION
■
A novel, scalable process for the synthesis of 2′-C-
methylguanosine-5′-[2-[(3-hydroxy-2,2-dimethyl-1-oxopropyl)-
thio]ethyl-N-benzylphosphoramidate] (IDX184, 1) via the use
of mild, PBA transient protection was developed. The physical
and chemical characteristics of 1 presented a variety of
challenges, notably its preferred solubility in organic/aqueous
solvent mixtures and sensitivity to acidic and basic conditions.
H
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
under argon. The solids were collected by filtration under
reduced pressure and dried under vacuum between 30 and 35
°C to afford 999 mg (77%) of 2′-C-methylguanosine-2′,3′-O-
phenylboronate 11 as a white solid: mp 282−284 °C dec; m/z
was removed by distillation under vacuum with a maximum
jacket temperature of 33 °C. A second charge of MTBE (118
L) was performed, and a further 88.7 kg of distillate was
recovered. Distillation ceased, and a third charge of MTBE
(471 L) was made prior to cooling the mixture to 21 °C. After
7 h, the diastereomer ratio by 31P NMR was 0.459:0.540. After
32 h, the diastereomer ratio was 0.476:0.524. The solids were
recovered by filtration, and the wet cake was washed with
MTBE (64 L) and EtOH (64 L) before the filter was purged
with nitrogen. Drying under vacuum at 45 °C with a nitrogen
purge gave trityl-protected phosphoramidate 6 as a white solid
(20.7 kg, 85% purity adjusted yield, 97.7% LCAP).
(ESI−, direct injection) 382.0 [M−H]− 100%; νmax (KBr disc,
20
cm−1) 3575, 3405, 3196 (br), 1690, 1638, 1604, 1354; [α]D
=
20
−27.49 (c 1.0, DMSO containing 0.1% H2O) at t = 0 h; [α]D
1
= −2.31 (c 1.0, DMSO containing 0.1% H2O) at t = 28 h; H
NMR (400 MHz, DMSO-d6) 1.23 (3H, s, CH3), 3.78 (1H, ddd,
J5′,5″ = 12.1 Hz, J5′,OH‑5′ = 6.1 Hz, J5′,4′ = 4.4 Hz, H-5′), 3.84
(1H, ddd, J5″,5′ = 12.1 Hz, J5″,OH‑5′ = 5.1 Hz, J5″,4′ = 3.6 Hz, H-
5″), 4.15 (1H, m, H-4′), 4.83 (1H, d, J3′,4′ = 4.7 Hz, H-3′), 5.24
(1H, a-t, JOH‑5′,5′/5″ = 5.6 Hz, OH-5′), 6.13 (1H, br-s, H-1′),
6.55 (2H, s, NH2), 7.45 (2H, t, J = 7.2 Hz, 2 × Ar-Hmeta), 7.57
(1H, dt, J = 7.2 Hz, J = 1.3 Hz, Ar-Hpara), 7.77 (2H, dd, J = 7.2
Hz, J = 1.3 Hz, 2 × Ar-Hortho), 8.00 (1H, s, H-8), 10.71 (1H, s,
N-H); 13C NMR (100 MHz, DMSO-d6) 20.28 (CH3), 60.90
(C-5′), 85.24 (C-3′), 85.79 (C-4′), 91.02, 91.04 (C-1′, C-2′),
116.35, 150.72, 153.75, 156.62 (C-2, C-4, C-5, C-6), 127.94 (2
× Ar-Cmeta), 131.84 (Ar-Cpara), 134.56 (2 × Ar-Cortho), 135.57
(C-8) (Ar-Cipso not observed). Anal. Calcd for C17H18N5O5B:
C, 53.29; H, 4.73; N, 18.28; B, 2.82. Found: C, 53.53; H, 4.74;
N, 18.01; B, 2.74.
A second reaction using 8.0 kg of 3 was processed similarly
to provide additional trityl-protected phosphoramidate 6 (21.7
kg, 85% purity adjusted yield, 97.0% LCAP, diastereomer ratio
0.457:0.543): m/z (ESI+) 869.40 [M+H]+ 100%; m/z (ESI−)
1
867.46 [M−H]− 100%; H NMR (400 MHz, DMSO-d6) two
diastereomers 0.83 (3H, s, CH3), 1.13 (6H, s, (CH3)2C), 3.00−
3.11 (4H, m, CH2S, CH2OTr), 3.83−4.01 (6H, m, H-3′, H-4′,
CH2O, CH2Ph), 4.12−4.28 (2H, m, H-5′, H-5″), 5.16 (1H, s,
OH-2′), 5.39, 5.41 (2 × 0.5H, 2d, 2 × J = 6.4 Hz, OH-3′), 5.65
(1H, m, P-N-H), 5.77 (1H, s, H-1′), 6.52 (2H, br-s, NH2),
7.17−7.43 (20H, m, 20 × Ar-H), 7.77, 7.78 (2 × 0.5H, 2s, H-8),
10.40 (1H, br-s, N-H).
2′-C-Methylguanosine-5′-[2-[(3-trityloxy-2,2-dimeth-
yl-1-oxopropyl)thio]ethyl-N-benzylphosphoramidate]
(6). 2′-C-Methylguanosine 3 (8.0 kg, 26.91 mol, 1.00 equiv),
anhydrous sodium sulfate (9.60 kg, 67.59 mol, 2.51 equiv), and
PBA (3.44 kg, 28.21 mol, 1.05 equiv) were charged to a reactor.
MeCN (48.9 L) was charged to the reactor, and agitation
commenced (100 rpm). The batch was heated to reflux (80
2′-C-Methylguanosine-5′-[2-[(3-hydroxy-2,2-dimeth-
yl-1-oxopropyl)thio]ethyl-N-benzylphosphoramidate]
(1). Trityl-protected phosphoramidate 6 (20.7 kg, assumed
23.82 mol, 1.00 equiv), MeCN (166 L), and EtOH (166 L)
were charged to a reactor, and agitation commenced (100
rpm). In a separate vessel, a dry HCl solution was prepared by
charging acetyl chloride (2.81 kg, 35.80 mol, 1.50 equiv) to a
mixture of EtOH (20.5 L) and MeCN (20.5 L) over 0.5 h,
maintaining internal temperature <35 °C. A line rinse with
MeCN (3.9 L) was added to the vessel and stirred for 0.5 h.
The HCl solution (37.5 kg) was charged to the reactor over 8
min and rinsed forward with MeCN (6.4 L). The reaction was
held at 25 °C for 11.5 h, at which point analysis by HPLC
indicated the LCAP ratio of 1 to 6 was 11.3:1. Dowex 66 basic
resin (45 kg) was washed with a mixture of MeCN (150 L) and
water (150 L). The washed resin (35.2 kg) was charged to the
reactor over 8 min, during which time no temperature change
was observed. After 1 h, the solution pH was 4. Additional
Dowex 66 resin (3.2 kg) was added, and after a further 0.5 h,
the measured pH was 6. The Dowex resin was removed by
filtration and washed with a mixture of MeCN (43 L) and
water (43 L). A clear, colorless solution of crude 1 was obtained
(392 kg, 88.1% LCAP).
Solvent (351 kg) was partially removed by distillation under
vacuum with a maximum jacket temperature of 40 °C. Water
(414 L) and MeCN (621 L) were charged to the concentrate,
and a solution was obtained on stirring at 100 rpm for 1.5 h at
42 °C. After the mixture cooled to 26 °C, toluene (357 L) was
charged. A light slurry was observed, at which point additional
water (45 L) and MeCN (57 L) were added. Dissolution was
complete after a further 1.5 h, at which point the phases were
allowed to separate at 25 °C for 1 h. The organic phase (733
kg) was discarded. Analysis of the aqueous phase (598 kg) by
HPLC indicated 97.7% LCAP with a 0.465:0.535 diastereomer
ratio. The aqueous phase was charged with MeCN (164 L) and
toluene (207 L), and the mixture was agitated at 100 rpm for
0.25 h. The phases were allowed to separate at 25 °C for 1 h.
The organic phase (340 kg) was discarded. Analysis of the
aqueous phase (564 kg) by HPLC indicated 97.9% LCAP with
1
°C) and held for 4.5 h. Analysis by H NMR indicated 99%
conversion to the desired phenylboronate 11, and the batch
was cooled to 22 °C. Triethylammonium 2-(2,2-dimethyl-3-
(trityloxy)propanoylthio)ethyl phosphonate33 4 (22.60 kg, 88%
LCAP, 33.97 mol, 1.26 equiv) was dissolved in MeCN (17.6 L)
and charged to the reactor. After rinsing forward with MeCN
(6.4 L), the batch was cooled to −2 °C. In a separate vessel, a
mixture of pyridine (22.96 L, 283 mol, 10.55 equiv) and
trimethylacetyl chloride (16.38 kg, 135.84 mol, 5.05 equiv) was
prepared. This mixture was charged over 2.5 h to the batch,
maintaining an internal temperature <8 °C. The charge vessel
was rinsed forward with MeCN (12.8 L), and the batch was
warmed to 10 °C over 1.5 h and then held between 10 and 17
°C. After 1 h, analysis by HPLC indicated the LCAP ratio of P-
H intermediate to 3 was 7.65:1. The reaction mixture was
diluted with MeCN (25.2 L) and cooled to −3 °C.
Benzylamine (42.00 kg, 391.97 mol, 14.56 equiv) was charged
over 1.75 h, maintaining the batch temperature <8 °C, followed
with a MeCN (6.5 L) rinse. Carbon tetrachloride (48.27 kg,
313.81 mol, 11.7 equiv) was added over 1.75 h, maintaining the
batch temperature <10 °C, followed with a MeCN (6.4 L)
rinse. After 2.5 h, analysis by HPLC indicated 0.60% LCAP P-H
intermediate remaining. MTBE (240 L) was charged, and the
batch was warmed to 25 °C. Aqueous citric acid solution
(21.3% w/w, 376.1 kg, 15.5 equiv) was added, and cleavage of
the cyclic boronate was effected over 1 h. The phases were
allowed to separate for 3 h. The lower aqueous phase was
discarded. Additional MeCN (6.2 L) was added, and the
organics were washed with an aqueous solution (236.0 kg) of
NaHCO3 (3.4% w/w) and NaCl (8.4% w/w). The batch was
agitated for 0.25 h before the phases were allowed to separate
over 2.5 h. The lower aqueous phase was discarded. MTBE
(118 L) was charged to the organics, and 118.7 kg of solvent
I
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
a 0.465:0.535 diastereomer ratio. Solvent (305 kg) was partially
removed from the aqueous phase by distillation under vacuum
with a maximum jacket temperature of 45 °C. MeCN (202 L)
and EtOAc (574 L) were charged to the concentrate, and the
mixture was agitated (100 rpm) for 0.25 h. The phases were
allowed to separate at 25 °C for 4.5 h. Analysis of the organic
phase (699 kg) by HPLC indicated 98.9% LCAP with a
0.466:0.534 diastereomer ratio. MeCN (202 L) and EtOAc
(574 L) were charged to the aqueous phase (216 kg), and the
mixture was agitated (100 rpm) for 2 h. The phases were
allowed to separate at 25 °C for 4 h. The second aqueous phase
(147 kg) was discarded. Analysis of the second organic phase
(744 kg) by HPLC indicated 98.5% LCAP with a 0.452:0.548
diastereomer ratio. Combination of the two organic phases
provided a total product solution of 1443 kg.
A second deprotection reaction of 6 (21.7 kg, assumed 24.97
mol, 1.00 equiv) was performed in a similar manner: analysis of
the first organic product phase (748.5 kg) by HPLC indicated
98.8% LCAP with a 0.464:0.536 diastereomer ratio, and
analysis of the second organic product phase (798.9 kg) by
HPLC indicated 98.5% LCAP with a 0.443:0.557 diastereomer
ratio. Combination of the two organic phases provided a total
product solution of 1547 kg.
two diastereomers 9.78, 9.92 (1P, 2s); HRMS (ESI) [M+H]+
m/z calcd for C25H36N6O9PS 627.2002, found 627.2010. Anal.
Calcd for C25H35N6O9PS: C, 47.92; H, 5.63; N, 13.41; P, 4.94.
Found: C, 48.18; H, 5.76; N, 13.30; P, 4.98.
ASSOCIATED CONTENT
* Supporting Information
■
S
Crystallographic information for the X-ray structure determi-
nation of the Sp diastereomer of IDX184 and experimental
details for the formation of phosphoramidate 6 using DCDMH.
This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: mayes.ben@idenix.com.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Administrative and operational support provided by Robert
Mayers and Ricardo Barrios is highly appreciated. Ricerca
Biosciences are acknowledged for performing the production of
1 in their facility and SAFC Pharmorphix for the X-ray
structure determination of the Sp diastereomer.
The two product solutions (combined 2990 kg, average
98.7% LCAP) were charged into the same reactor, and solvent
(2868 kg) was partially removed by distillation under vacuum
with a maximum jacket temperature of 40 °C. 2-Propanol (115
L) was charged to the concentrate, and solvent (90 kg) was
partially removed by distillation under vacuum with a maximum
jacket temperature of 40 °C. This 2-propanol/distillation
process was repeated four more times, removing a total of
438 kg of distillate. Isopropyl acetate (93 L) was charged to the
concentrate, and solvent (82 kg) was partially removed by
distillation under vacuum with a maximum jacket temperature
of 40 °C. This isopropyl acetate/distillation process was
repeated six more times, removing a total of 558 kg of distillate.
An eighth charge of isopropyl acetate (103 L) was performed,
and the batch was held at 35 °C for 0.75 h. The slurry was
cooled to 0 °C over 4 h and held at 0 °C for 2 h. The solids
were filtered under vacuum and washed with isopropyl acetate
(77 L), and the filter was purged with nitrogen. Drying under
vacuum at 45 °C with a nitrogen purge gave phosphoramidate
1 as a white solid (20.2 kg, 69% purity adjusted yield, 99.4%
LCAP34 with a 0.457:0.543 diastereomer ratio): m/z (ESI+)
627.05 [M + H]+ 100%, 1253.55 [2M+H]+ 20%; νmax (KBr
REFERENCES
■
(1) Lalezari, J.; Asmuth, D.; Casiro, A.; Vargas, H.; Lawrence, S.;
Dubuc-Patrick, G.; Chen, J.; McCarville, J.; Pietropaolo, K.; Zhou, X.-
J.; Sullivan-Bolyai, J.; Mayers, D. Antimicrob. Agents Chemother. 2012,
56 (12), 6372−6378.
(2) Zhou, X.-J.; Pietropaolo, K.; Chen, J.; Khan, S.; Sullivan-Bolyai, J.;
Mayers, D. Antimicrob. Agents Chemother. 2011, 55 (1), 76−81.
(3) (a) Sofia, M. J.; Chang, W.; Furman, P. A.; Mosley, R. T.; Ross, B.
S. J. Med. Chem. 2012, 55, 2481−2531. (b) Feeney, E. R.; Chung, R. T.
BMJ 2014, 348, g3308. (c) De Clercq, E. Biochem. Pharmacol. 2014,
89, 441−442. (d) Wendt, A.; Adhoute, X.; Castellani, P.; Oules, V.;
Ansaldi, C.; Benali, S.; Bourliere, M. Clin. Pharmacol. 2014, 6, 1−17.
(e) Caillet-Saguy, C.; Lim, S. P.; Shi, P.-Y.; Lescar, J.; Bressanelli, S.
Antiviral Res. 2014, 105, 8−16.
(4) For example: solubility of 1 in water = 0.008 mol L−1; in EtOH =
0.04 mol L−1; in water/EtOH 1:1 = 0.16 mol L−1.
(5) Sommadossi, J.-P.; Gosselin, G.; Pierra, C.; Perigaud, C.;
Peyrottes, S. Int. Patent WO2008082601, 2008.
(6) Ross, B. S.; Sofia, M. J.; Pamulapati, G. R.; Rachakonda, S.;
Zhang, H.-R. Int. Patent WO2011123668, 2011.
1
disc, cm−1) 3334 (br), 1685, 1635, 1598, 1571; H NMR (400
(7) Mayes, B. A.; Arumugasamy, J.; Baloglu, E.; Bauer, D.; Becker, A.;
Chaudhuri, N.; Latham, G. M.; Li, J.; Mathieu, S.; McGarry, F. P.;
Rosinovsky, E.; Stewart, A.; Trochet, C.; Wang, J.; Moussa, A. Org.
Process Res. Dev. 2014, 18, 717−724.
(8) Discussion of other methodologies to couple nucleosides to form
phosphoramidate prodrugs is provided in ref 7. For a full account, see:
Padere, U.; Garnier-Amblard, E. C.; Coats, S. J.; Amblard, F.; Schinazi,
R. F. Chem. Rev. 2014, 114, 9154−9218.
(9) (a) Atherton, F. R.; Openshaw, H. T.; Todd, A. R. J. Chem. Soc.
1945, 660−663. (b) Georgiev, E. M.; Kaneti, J.; Troev, K.; Roundhill,
D. M. J. Am. Chem. Soc. 1993, 115, 10964−10973.
(10) The acceptable range for the 1:1 diastereomer specification was
a Rp:Sp ratio of 45:55 to 55:45.
(11) Liquid chromatography area percentage.
(12) THF/water/saturated aqueous NaHCO3 vol ratio was
1:0.25:0.25.
MHz, DMSO-d6) two diastereomers 0.85 (3H, s, CH3), 1.11
(6H, s, (CH3)2C), 3.05 (2H, m, CH2S), 3.44 (2H, d, J = 5.5 Hz,
CH2OH), 3.89−4.02 (6H, m, H-3′, H-4′, CH2O, CH2Ph),
4.14−4.20 (1H, m, H-5′), 4.22−4.30 (1H, m, H-5″), 4.97 (1H,
t, J = 5.5 Hz, CH2OH), 5.22 (1H, s, OH-2′), 5.43, 5.46 (2 ×
0.5H, 2d, 2 × J = 6.4 Hz, OH-3′), 5.68 (1H, m, P-N-H), 5.78
(1H, s, H-1′), 6.56 (2H, br-s, NH2), 7.20−7.24 (1H, m, Ar-H),
7.28−7.34 (4H, m, 4 × Ar-H), 7.80, 7.81 (2 × 0.5H, 2s, H-8),
10.69 (1H, br-s, N-H); 13C NMR (100 MHz, DMSO-d6) two
diastereomers 19.35 (CH3), 21.27 (C(CH3)2), 27.63, 27.70
(CH2S), 43.65, 43.69 (PhCH2), 51.11 (C(CH3)2), 63.24, 63.28
(CH2O), 64.41, 64.71 (C-5′), 67.78 (CH2OH), 72.01, 72.21
(C-3′), 77.61, 77.63 (C-2′), 79.53, 79.61 (C-4′), 89.81, 89.90
(C-1′), 115.87, 115.91 (C-5), 126.14 (Ar-Cpara), 126.60, 126.61,
127.55, 127.56 (4 × Ar-C), 134.42, 134.58 (C-8), 139.82,
139.88 (Ar-Cipso), 150.19, 150.22 (C-4), 153.07, 153.09, 156.18
(C-2, C-6), 203.39 (COS); 31P NMR (162 MHz, DMSO-d6)
(13) A typical elution profile consisted of 1.5% MeCN/water, 21 vol;
5% MeCN/water, 16 vol; 10% MeCN/water, 35 vol; 15% MeCN/
water, 35 vol; 18% MeCN/water, 33 vol; 20% MeCN/water, 33 vol;
25% MeCN/water, 118 vol.
J
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(14) ES was quantified by a validated headspace gas chromatographic
analytical method.
(15) Vide infra.
(16) At three ICH storage conditions: 5 °C; 25 °C /40% relative
humidity (RH); 40 °C /75% RH.
(17) World Health Organization International Agency for Research
on Cancer. Monographs on the Evaluation of Carcinogenic Risks to
Humans, 1976, Vol. 11, p 257 and 1987, Suppl. 7, p 63.
(18) FDA Center for Drug Evaluation and Research. Genotoxic and
Carcinogenic Impurities on Drug Substances and Products: Recom-
mended Approaches, Guidance for Industry (Draft), 2008.
(19) European Medicines Agency, Committee for Medicinal
Products for Human Use. Guideline on the Limits of Genotoxic
Impurities, 2006.
(20) International Conference on Harmonization. Assessment and
Control of DNA Reactive (Mutagenic) Impurities in Pharmaceuticals
to Limit Potential Genotoxic Risk, Harmonized Tripartite Guideline
M7, Step 4 version, 2014.
(21) Impurity 13 was qualified in preclinical toxicology studies
allowing this specification to be set for the final API.
(22) The cyclic boronates were completely hydrolyzed under the
HPLC conditions; therefore, the respective deprotected species were
monitored. Conversion of 7:1 at this stage was typical, and addition of
further PivCl/pyridine did not improve isolated yields.
(23) LCAP represents the sum of both diastereomers of P-OEt 13.
(24) The first and second organic layers contained a 10:1 ratio of
product 1 content.
(25) The extracts contained EtOAc, MeCN, and ∼7 wt% water.
(26) As per: International Conference on Harmonization.
Impurities: Guideline for Residual Solvents, Harmonized Tripartite
Guideline ICH Q3C (R5), Step 4 version, 2011.
(27) Based on the categorization of CCl4 as a Class 1 solvent under
ICH Q3C.
(28) O’Donovan, M. R.; Mee, C. D.; Fenner, S.; Teasdale, A.;
Phillips, D. H. Mutat. Res. 2011, 724, 1−6.
(29) Methods to control residual arylboronic acids in APIs are
discussed in http://www.andersonsprocesssolutions.com/controlling-
residual-arylboronic-acids-as-potential-genotoxic-impurities-in-apis/
#_ednref1 (accessed 17 Nov 2014).
(30) (a) Baek, K.-W.; Song, S.-H.; Kang, S.-H.; Rhee, Y.-W.; Lee, C.-
S.; Lee, B.-J.; Hudson, S.; Hwang, T.-S. J. Ind. Eng. Chem. 2007, 13,
452−456. (b) http://www.dow.com/assets/attachments/business/
ier/ier_for_industrial_water_treatment/amberlite_ira743/tds/
amberlite_ira743.pdf (accessed 17 Nov 2014).
(31) Based on a 200 mg daily dose of 1, 6 ppm equated to 1.2 μg,
which was 16-fold below the 20 μg per day limit for PGIs based on a
1−3 month clinical trial duration.
(32) Jimenez-Gonzalez, C.; Ponder, C. S.; Broxterman, Q. B.;
Manley, J. B. Org. Process Res. Dev. 2011, 15, 912−917.
(33) Vide supra, see ref 7.
(34) 99.4% LCAP integrating all impurities ≥0.05%.
K
DOI: 10.1021/op5003837
Org. Process Res. Dev. XXXX, XXX, XXX−XXX