Organic Process Research & Development
Communication
hydrogen-bonding through the chloride ion, creating hexagonal
tunnels down the c-axis capable of being occupied by solvent
molecules, typically 2-PrOH, which is the solvent from which
this salt is isolated. In a practical sense, this means that the
isolated IPI-926 drug substance will therefore contain levels of
2-PrOH that are significantly higher than ICH limits of 5000
ppm:17,18 The theoretical amount of 2-PrOH for the
stoichiometric solvate would be 10% or 100 000 ppm, and
typical values observed ranged from 80 000 ppm to 100 000
ppm. However, these levels were deemed acceptable for this
drug substance, as the drug product manufacture employed a
low shear wet granulation, which displaced the 2-PrOH with
water. The resulting drug product contained typically 50−200
ppm of 2-PrOH.
smaller particles and regrowth from existing crystals. The result
of the heating cycle was therefore a product with higher purity
and a more uniform particle size that ultimately offered an
improved filterability/processability. The length of the heat
cycle was probed but no adverse consequence (yield or purity
of the isolated solid) was observed from heating the suspension
at reflux for up to 35 h. Specifically, the formation of 2-
chloropropane20 was investigated, and no clear effect on the
content of this impurity in the isolated DS was observed from
this heat cycle. Should 2-chloropropane be forming as a result
of this reaction, it appears to neither be trapped in the isolated
drug substance nor result in formation of other impurities,
including the N-alkylation product, in appreciable levels.
Since the debenzylation reaction was studied separately from
the salt formation, the two processes had to be decoupled.21 In
order to do so and to normalize the equivalents of HCl used for
the salting and crystallization of IPI-926, it became important to
assay the amount of IPI-926 in solution prior to the
crystallization and to properly titer the HCl equivalents used
for the crystallization of IPI-926·HCl. Operationally, since the
IPI-926 solution obtained after the scavenging operation is in a
mixture of 2-MeTHF and 2-PrOH, a distillation is performed to
remove water and exchange the solvent to mostly 2-PrOH prior
to the salting. The effect of 2-MeTHF on the yield of the salting
and purity of IPI-926 was investigated. To this end, no
significant difference in purity or yield was observed for any of
the doped samples ranging from 0 to 20 000 ppm residual 2-
MeTHF. Typical 2-MeTHF content in the 2-PrOH solution of
IPI-926 used in the crystallization was shown to be <100 ppm.
The assay of the IPI-926 solution is therefore performed on the
2-PrOH solution prior to salting.
Salting DoE. Finally, a DoE approach was taken to fully
understand the performance of the IPI-926 salting procedure.
The amount of seed used was increased from 1.5% and fixed at
5% based on the solubility22 of IPI-926·HCl in 2-PrOH and to
ensure that the seed would not dissolve completely in the
reaction mixture. An I-Optimal design was chosen to investigate
the following variables: (1) the maximum cycle temperature
(the length of the cycle was fixed at 6 h; 25 °C, 55 °C, ∼84 °C
(reflux)), (2) the equivalents of HCl relative to IPI-926 (1.0,
1.15, 1.3 equiv), (3) the addition rate of HCl (fast, moderate,
slow; corresponding to an addition time of 0, 1, and 2 h,
respectively23), and (4) the final aging temperature (0 °C, 12.5
°C, 25 °C). These variables were combined into a custom I-
optimal design and resulted in a study design comprising 16
experiments. The responses monitored were (1) product yield,
(2) purity, and (3) mass balance including the mother liquors
(a representation of any decomposition or side-reaction that
could have occurred). All of the study was completed from a
single lot of IPI-926 solution in 2-PrOH, which was partitioned
and assayed after distillation to ensure a uniform starting
material for all salting experiments.
IPI-926·HCl Salt Formation and Isolation. The original
conditions used for the synthesis of IPI-926·HCl called for
removing and salting 1.5% of the batch for use as a seed in the
main batch. The seed was added after an initial addition of the
HCl to the batch (prior to nucleation), and then the remaining
HCl solution (in 2-PrOH) was added over the course of 7−8 h.
This approach addressed the question of an appropriate seed
for manufacturing under cGMP (i.e., whether a seed needs to
be generated under cGMP or can be treated as a raw material
with the proper release controls) and took a conservative
approach to dealing with this issue. Moreover, since the impact
of seed type (i.e., crystal size, habit, etc.) had not been
established, this approach allowed for the most flexibility during
production. However, upon review of this process, a number of
limitations were identified in this salting procedure: (1) the
quantity of the seed was small, and set arbitrarily (without
understanding of the IPI-926·HCl solubility); (2) the addition
rate of the HCl was very slow (took a very long time in
production); and (3) the procedure as a whole was very
complicated operationally. Additionally, the stoichiometry of
the HCl used for this salting had been arbitrarily set to 1.3
equiv relative to the starting material, compound 1, with no
clear understanding of the effect of excess HCl on the purity of
the isolated material or on the actual yield of the debenzylation
reaction. For these reasons, and in an effort to produce material
that was of consistently high purity, the procedure was revised.
Second Generation Investigation. After a number of
experiments probing the effect of temperature, HCl equivalents,
seed quantity, seed type (added as a solid or as a freshly
prepared suspension), and addition rate, the highest purity
material was obtained after multiple heat/cool cycles of the IPI-
926·HCl crystals in 2-PrOH (Figure 8). Additional experiments
demonstrated the benefit of the seed crystals in terms of
allowing for a controlled crystal growth rather than a sudden
precipitation of the salt due to improper control of super-
saturated IPI-926·HCl (data not shown). Combined, these
results suggested that both the addition of a heat/cool cycle and
the use of a seeding protocol would improve the quality of the
product.
Yield was most affected by the stoichiometry of the HCl
added (Figure 9), but only for the low charge, which implies
that a slight excess of HCl is necessary to ensure maximum
recovery. The purity was most affected by the rate of HCl
addition (Figure 10), with the most pronounced effect being
observed when the addition rate was slowed down from “fast”
to “moderate”. The lower purity can be explained by the fact
that a faster addition rate of HCl will result in an uncontrolled
crystallization of the HCl salt and likely result in trapping of
impurities during the salt formation. Finally, mass balance was
also affected by the rate of HCl addition (Figure 11). This last
The behavior of the IPI-926·HCl crystals in suspension was
investigated at room temperature and during reflux using in-line
tools such as focused beam reflectance measurement (FBRM)
and particle vision and measurement (PVM) and off-line tools
such as polarized light microscopy (PLM). It was hypothesized
that the heat cycle broke up large agglomerations resulting from
uncontrolled nucleation/growth that could trap impurities
(initial nucleation results in acicular crystals) and subsequently
led to secondary crystal growth (as suggested by the appearance
of the stacked crystals in Figure 8) through dissolution of the
F
Org. Process Res. Dev. XXXX, XXX, XXX−XXX