Optimization and process improvement for LCZ696 by employing quality by design (QbD) principles

Article abstract of DOI:10.1016/j.tet.2020.131558

Efforts toward optimization and improvement for the synthesis of LCZ696 employing design of experiment (DoE) principles are described. By increasing the purity of intermediates and mitigating impurity risk during each step, a telescoped process was developed via removal of isolation of intermediates with the overall yield increased by 28.5% from 45.3% to 73.8%. And the whole production cycle was also shortened from 12 days to 7 days with simplified operations and restored process greenness. Meanwhile, the corresponding impurity profile was thus studied in detail and well documented.

Full text of DOI:10.1016/j.tet.2020.131558

Tetrahedron xxx (xxxx) xxx
Contents lists available at ScienceDirect
Tetrahedron
journal homepage: www.elsevier.com/locate/tet
Optimization and process improvement for LCZ696 by employing
quality by design (QbD) principles
,
d
,
,
d
,
,
,
, d
Zhijun Chen a 1, Hailong Wang b 1, Shuming Wu b d, Jian Wang b d, Chenxia Zhang b
,
Hua Yang a **, Zhongqing Wang b
,
, c, d, *
a College of Chemistry and Chemical Engineering, Central South University, Changsha, 410083, PR China
b State Key Laboratory of Anti-Infective Drug Development, Sunshine Lake Pharma Co., Ltd., Dongguan, 523871, PR China
c School of Pharmacy, Xiangnan University, Chenzhou, 423000, Hunan, China
d HEC Research and Development Center, HEC Pharm Group, Dongguan, 523871, PR China
a r t i c l e i n f o
a b s t r a c t
Article history:
Efforts toward optimization and improvement for the synthesis of LCZ696 employing design of exper-
iment (DoE) principles are described. By increasing the purity of intermediates and mitigating impurity
risk during each step, a telescoped process was developed via removal of isolation of intermediates with
the overall yield increased by 28.5% from 45.3% to 73.8%. And the whole production cycle was also
shortened from 12 days to 7 days with simplified operations and restored process greenness. Meanwhile,
the corresponding impurity profile was thus studied in detail and well documented.
© 2020 Published by Elsevier Ltd.
Received 29 July 2020
Received in revised form
24 August 2020
Accepted 28 August 2020
Available online xxx
Keywords:
LCZ696
DoE
Process
Optimization
1. Introduction
prepared by combining sacubitril [3,4b,5], sodium sacubitril [4a],
calcium sacubitril [3b,3c,5a,7] or sacubitril dicyclohexylamine salt
LCZ696 (Fig. 1) is the first example of a dual-acting pharma-
ceutical built as a supramolecular complex delivering two phar-
macologic effectsdangiotensin receptor 1 (AT1) blockage and
neprilysin (NEP) inhibition [1]. It is a fixed-dose combination with
sacubitril and valsartan (VST) [1] and marketed by Novartis under
the brand name Entresto®, which is indicated to reduce the risk of
cardiovascular death and hospitalization for heart failure in pa-
tients with chronic heart failure and reduce ejection fraction. And
LCZ696 was approved under the FDA’s priority review process for
the use in heart failure on July 7, 2015 [2].
[8] with valsartan [3,4b,5] or sodium valsartan [4a,5b]. Except for
Merck’s newly disclosed synthesis for a sacubitril intermediate
employing flow chemistry [9], there are two typical synthetic
routes to sacubitril (Scheme 1). The first route [3] adopted a strat-
egy that using a pre-constructed biaryl compound 1 as a starting
material to perform asymmetric hydrogenation in order to intro-
duce the second stereocenter of the molecule, while the second
route [6] tended to build the biaryl moiety by a coupling reaction
after the two stereocenters having been set up. Route 1 was
preferred as the extra palladium-catalyzed coupling step of route 2
would definitely add to the cost and environmental footprints for
the whole process and possibly lead to a loss of optical purity and
yield, albeit both of the two routes suffered from several minor
drawbacks brought by isolation (extraction, filtration, etc.) of in-
termediates for each step, such as high process complexity, long
processing time and usage of multi-solvent.
As a combination of sacubitril and valsartan, the preparation
method was to combine the two components which were in
different forms [3e8]. In reported prior processes, LCZ696 was
* Corresponding author. State Key Laboratory of Anti-Infective Drug Develop-
ment, Sunshine Lake Pharma Co., Ltd., Dongguan, 523871, PR China.
** Corresponding author. College of Chemistry and Chemical Engineering, Central
South University, Changsha, 410083, PR China.
To address the aforementioned issues, we envisioned devel-
oping a telescoped and optimized process based on route 1. Since
the initial introduction of Quality by Design (QbD) and the frame-
work outlined by various authorities, such as the International
Council for Harmonisation of Technical Requirements for
E-mail addresses: Hyangchem@csu.edu.cn (H. Yang), Wangzhongqing@hec.cn
(Z. Wang).
1
These authors contributed equally.
https://doi.org/10.1016/j.tet.2020.131558
0040-4020/© 2020 Published by Elsevier Ltd.
Please cite this article as: Z. Chen et al., Optimization and process improvement for LCZ696 by employing quality by design (QbD) principles,
Tetrahedron, https://doi.org/10.1016/j.tet.2020.131558

2
Z. Chen et al. / Tetrahedron xxx (xxxx) xxx
stereocenter to the molecule, which was one of the major trans-
formations in the workflow. Synthetic methodologies for preparing
compound 2 were well afforded by literatures [3,13beh], among
which asymmetric hydrogenation with transition metal and chiral
ligands was the most common one. [3b,13b,eeh] Although there
were many other combinations of metal/ligand complex [13eeh],
mixed catalyst system involving diiodo (p-cymene)ruthenium (II)
dimer (LCZ-A) and MandyPhossl-M004-1 (LCZ-B) seemed to pro-
vide the ideal selectivity for the conversion of (R,E)-5-([1,10-
biphenyl]-4-yl)-4-((tert-butoxycarbonyl)amino)- 2-methylpent-2-
enoic acid (1) to its corresponding chiral carboxylic acid (2).
However, a key impurity in this asymmetric hydrogenation, dia-
stereomer of the product 2 (Scheme 3), which would finally lead to
the presence of impurity A in API was inevitable. Therefore, we
investigated reaction pressure and temperature through screening
experiments (see supporting information), but the two parameters
did not significantly impact the formation of the isomer. Even un-
der our optimized conditions (3.0 MPa, 65 ꢀC), the content of dia-
stereomer in 2 was still around 2%.
Fig. 1. The chemical structure of LCZ696.
Pharmaceuticals for Human Use (ICH) [10], the Food and Drug
Administration (FDA) [11a], and the European Medicines Agency
(EMA) [11b], the pharmaceutical industry has displayed a keen
interest replacing the time-consuming traditional optimization
approach, varying one factor/variable at a time (OFAT, also known
as OVAT) into the multivariate design of experiments (DoE) [12].
Based on the statistical design and analysis of experiments, the
ranges of critical material attributes (CMAs) and critical process
parameters (CPPs) are identified to achieve products with critical
quality attributes (CQAs), together defining a broader operational
space compared to the narrow one established by self-limited OFAT,
which has also driven down the risk to quality in the production of
active pharmaceutical ingredients (APIs) [12a].
In the original process, the diastereomer impurity was
completely removed in the isolation and purification of interme-
diate 2 [3b,13b]. As such, it could not impact the quality of API.
However, when the process is telescoped, the isomer impurity will
be carried through the sequential steps until finally get purged. To
our delight, a rejection of the isomer impurity was observed in later
stage of the process.
Herein we describe an efficient telescoped route with the aid of
DoE methodology with only one intermediate isolated (4c, Scheme
2) while at least three intermediates were involved in the first re-
ported route. With broad and robust operational spaces estab-
lished, this telescoped process considerably increases the yield by
28.5% from 45.3% [3b,3c] to 73.8% with a purity of the final product
over 99.9% and significantly shortened the production cycle by 42%
from 12 days to 7 days by the removal of isolation of two in-
termediates. The DoE approach also enabled a more thorough un-
derstanding of the process, a compressed development timeline
and a reduction in overall cost.
2.2. Process development for 3
With a crude solution of intermediate 2 in hand, we focused on
the next step in the workflow. For a successful transformation into
compound 3, both a deprotection of the Boc group on the primary
amine and an ethyl esterification of the carboxylic group were
required for intermediate 2. Thus, the main byproduct in this step is
impurity 7, which had gone through an N-Boc deprotection but
failed to finish the esterification. A screening experiment was
designed to investigate the impact of different kinds of acids and
reaction duration to the formation of impurity 7 (Table 1). However,
as shown in Table 1 (entries 1e5), the formation of 7 was greatly
impacted by which acid was used but not influenced by the reac-
tion duration. Hence, thionyl chloride is still chosen as the optimal
reagent for the subsequent process optimization.
2. Results and discussion
2.1. Process development of 2
The first step of the process was to introduce the second
Base on the understanding of the reaction, the equivalents of
Scheme 1. Reported Approaches to the synthesis of LCZ696.
Please cite this article as: Z. Chen et al., Optimization and process improvement for LCZ696 by employing quality by design (QbD) principles,
Tetrahedron, https://doi.org/10.1016/j.tet.2020.131558

Z. Chen et al. / Tetrahedron xxx (xxxx) xxx
3
Scheme 2. The telescoped route of LCZ696.
Scheme 3. Diastereomer impurity in intermediate 2.
Table 1
Table 2
Acid and reaction time screen of compound 3a.
Three process parameters were considered for the optimization DoEa.
Entry
Acid screen
Time (h)
3 (%)b
7 (%)b
Symbol
Parameter
Unit
Low
High
1
2
3
4
5
6
HCl (conc.)
H2SO4 (conc.)
MeSO3H
TsOH
SOCl2
SOCl2
24
24
24
24
24
2
94.35
96.78
97.64
95.49
99.16
99.20
2.17
1.42
0.91
0.91
0.73
0.68
1
2
3
Temperature
SOCl2
EtOH
ꢀC
Equiv
volume
25
1.0
5
65
1.5
15
a
Central composite face-centered design (CCF) with two center points was
planned to study the effect of the three important input process parameters on the
response variables with all of the other parameters kept at predefined levels.
a
Reaction conditions: compound 2 (2.0 g), acid (1.5 equiv.), ethanol (40 mL),
65 ꢀC.
b
Area % by HPLC of crude compound.
From the analysis of variance (ANOVA, see Supporting Infor-
mation) and contour plots (Fig. 2) of response variables, we found
that the assay yield of 3 and the formation of impurity 7 were
directly affected by the equivalents of SOCl2 and the volumes of
EtOH while the temperature had a less significant impact. On the
basis of the results of the DoE, a higher equivalency of SOCl2 (>1.3
equiv) and a larger volume of EtOH resulted in an increase in yield
of 3 and a decrease of impurity 7, implying that the effect of EtOH as
a reactant outweighed that as a solvent. Additionally, the use of DoE
identified a secondary interaction between SOCl2 and EtOH, where
with high reaction concentration the effect of SOCl2 was amplified
(Fig. 2a). Under optimized conditions, a purity of over 98.8% was
achieved for compound 3 and less than 0.13% of impurity 7 was
formed. The model was validated by conducting three verification
SOCl2 and reaction temperature were identified as the potential
CPPs for the synthesis of 3. EtOH worked not only as a reactant in
the ethyl esterification but also as the solvent of the whole reaction.
Usually, a reaction was supposed to be promoted with an increased
charging of its reactant while suppressed by a lower reaction
concentration. Thus it was not straightforward to predict the
impact of EtOH on the reaction and it was also investigated. With
low and high ranges summarized in Table 2, a three-factor full
factorial DoE was performed to better understand the impact of
these parameters and to optimize the robustness of the reaction
(Supporting Information). Experimental designs were formulated
using the statistical software package Minitab 17 (Minitab LLC.).
Please cite this article as: Z. Chen et al., Optimization and process improvement for LCZ696 by employing quality by design (QbD) principles,
Tetrahedron, https://doi.org/10.1016/j.tet.2020.131558

4
Z. Chen et al. / Tetrahedron xxx (xxxx) xxx
purity of 4a (see supporting information). Although all of the four
solvents gave acceptable conversions of 3, some of them suffered
from drawbacks on manipulation which hinder them from wider
application. For example, when DCM or toluene was employed as a
solvent of this reaction, severe emulsification would be encoun-
tered and the two phases were still difficult to separate even left to
stand overnight at room temperature while EA and IPAc did not
have emulsification issues. However, EA was found to be hydro-
lyzed under basic conditions of the reaction (aqueous NaOH solu-
tion), making the pH adjustment of the mixture to 7.0e7.5 a bit
difficult. And IPAc was relatively stable under aqueous NaOH so-
lution due to greater steric hindrance. As a result, IPAc was
considered appropriate for this reaction.
With IPAc as solvent, the following tested factors were N,N-
Diisopropylethylamine (DIPEA, equiv), succinic anhydride (equiv)
and IPAc (volume) (Table 4). Based on an analysis of the data from
the DoE, the significant parameters influencing the reaction were
equivalents of succinic anhydride and volumes of IPAc (Fig. 3).
Interestingly, a peak purity of 4a (>98.0%) was observed under
higher concentrations (volumes of IPAc <12.5 vol) and with larger
amounts of succinic anhydride (1.2e1.5 equiv). As the reaction was
diluted, more succinic anhydride was required to maintain a high
assay yield; however, when the solvent volume was beyond
12.5 vol, no matter how much the succinic anhydride was charged,
the assay yield would only be less than 98.0%.
Under optimized conditions of 1.34 equiv of succinic anhydride,
1.2 equiv of DIPEA and 5 vol of IPAc, a purity around 98.0% for 4a
was observed. However, compared to its solid calcium salt 4c, 4a is
an oily intermediate which was not convenient for storage and
transportation. In addition, residual succinic anhydride passed on
the next step would also consume NaOH, attributing to a deficiency
in base and further leading to a decrease in yield of 5. Therefore, 4a
was transferred into 4c and was purified after a telescoped process
of three steps to achieve a purity of >99.0%.
2.4. Process development of 5 (crude product of LCZ696)
Fig. 2. Contour plots of the effect of equivalents of SOCl2 and ethanol volume on the
purity of 3 and content of impurity 7.
From a purification and manipulation convenience perspective,
we opted to use calcium sacubitril (4c) instead of free sacubitril (4a)
as the only isolated intermediate. Consequently, a dissociation step
was needed for 4c to achieve 4a, which was accomplished by
treatment of suspension of 4c in IPAc with diluted HCl aqueous
solution. After concentration and redissolution, 4a was combined
with VST under the presence of aqueous NaOH solution. After
precipitation and filtration, the crude LCZ696 was given as a solid
with a purity of over 99% by HPLC. Although acceptable purity was
experiments and the results were within the 95% confidence in-
terval (Table 3).
In prior arts [4a,5b,7], isolated compound 3 with a high purity
(>99.8%) was provided by a slurry along with a solvent swap, which
had mitigated part of the risk to quality of API. However, in the
telescoped process, the risk could only be driven down by a
fundamental understanding and a detailed optimization engen-
dered by DoE. With a purity of around 99%, unisolated compound 3
was used directly for the next step of functionalization.
Table 4
Three process parameters were considered for optimization DoE.a.
Entry
Parameter
Unit
Range
Low
2.3. Process development of 4c
high
For this subsequent amidation reaction, four alternative sol-
vents, namely ethyl acetate (EA), dichloromethane (DCM), toluene,
and isopropyl acetate (IPAc), were evaluated at first regarding the
1
2
3
DIPEA
Succinic anhydride
IPAc
Equiv
Equiv
Volume
1.1
1.0
5
1.5
1.5
15
a
All of the other parameters kept at predefined levels.
Table 3
Three verification experiments of 3.
Entry
SOCl2 (Equiv)
EtOH (volume)
Time (h)
3 (%)
7 (%)
0.05
0.05
0.06
purity of 3 95% CI(%)a
(98.71, 99.11)
7
1
2
3
1.5
1.5
1.5
15
15
15
24
24
24
99.00
98.95
98.95
95% CI(%)b
(0.0435, 0.1991)
a
Standard deviation of purity of 3 ¼ 0.0894%.
Standard deviation of purity of 7 ¼ 0.0357%.
b
Please cite this article as: Z. Chen et al., Optimization and process improvement for LCZ696 by employing quality by design (QbD) principles,
Tetrahedron, https://doi.org/10.1016/j.tet.2020.131558

Z. Chen et al. / Tetrahedron xxx (xxxx) xxx
5
Table 6
Three process parameters were considered for the DoE optimizationa.
Entry
Parameter
unit
Low
high
1
2
3
Acetone
VST
NaOH (aq.)
Volume
Equiv
Equiv
16
1.6
5.4
30
2.0
5.8
a
All of the other parameters kept at predefined levels.
Fig. 3. The contour plot of succinic anhydride and IPAC on the purity of 4a.
achieved, the yield of this step was not satisfying with only about
70%.
Therefore, a screening DoE was planned to determine the sig-
nificant factors which have great influences on the reaction yield
and product purity from the unimpactful ones (Table 5). A total of 9
experiments were run to investigate the impact of VST (equiva-
lents) and NaOH (equivalents and aqueous concentration). Based
on the results of screening experiments, the amount of aqueous
sodium hydroxide solution and VST charged were identified as
potential significant parameters. Based on the knowledge gained
from screening DoE, excess NaOH could cause a basic hydrolysis of
sacubitril, leading to the formation of Impurity D and thus render a
decrease in reaction yield due to product decomposition. Hence,
ranges for these two factors were redefined for a more detailed
evaluation on the impact of these two input variables on the yield
of crude LCZ696 (5) and the level of impurity D (Table 6). The
volume of acetone employed was also investigated, for the disso-
lution could affect the product yield or the impurity level.
From both the ANOVA analysis (see supporting information) and
the contour plot (Fig. 4), an interesting secondary interaction was
observed between the two mentioned input variables, equivalents
of charged NaOH and VST. In this case, when VST was charged less
than 1.85 equivalents, the yield of 5 decreased along with an
increasing addition of NaOH; when charged over 1.85 equivalents,
the yield had a positive correlation with the charged NaOH. As the
addition of aqueous NaOH solution increased, more water would
Fig. 4. The contour plot of equivalents of NaOH solution and VST on the yield of 5.
also be added into reaction system if the concentration remained
constant. As a result, a further dissolution of final product in excess
water could account for the aforementioned yield loss. On the other
hand, extra VST could cost more sodium hydroxide to form a stable
salt, leading to a base insufficiency for complex formation of
sacubitril and VST, and hence resulting in a drop in reaction yield by
charging less NaOH. As such, the proper range of amount of VST and
NaOH with peak reaction yield is within an irregular interval pre-
sented in Fig. 4.
Even though impurity 7 generated in the previous deprotection
step could likely result in a formation of impurity D through ami-
dation, it could be purged during the isolation of 4c with no addi-
tional risk to API quality. While in this step, as previously
mentioned, the level of impurity D was strongly affect by the
amount of NaOH employed due to a basic hydrolysis. Furthermore,
the insufficient charge of VST, where the sodium hydroxide is
relatively excessive, also rendered an increase in impurity D.
Table 5
Process understanding screen.a.
Entry
VST (equiv)
NaOH (equiv)
Concentration of NaOH (%)
HPLC (%)
Yield (%)
Impurity D
VST
4ab
1
2
3
4
5
6
7
8
9
1.84
1.88
1.92
1.96
2.00
1.80
1.80
1.80
1.80
5.15
5.26
5.37
5.48
5.60
5.04
5.04
5.04
5.04
46
46
46
46
46
35
40
46
50
0.02
0.01
0.05
0.02
0.01
0.06
NDd
NDd
0.01
36.62
36.81
36.94
36.64
36.76
36.58
37.34
36.44
36.93
63.33
63.16
62.96
63.32
63.21
63.20
62.49
62.67
62.87
73.7c
73.0
71.8
75.8
76.7
80.1
84.4
86.3
91.1
a
Reaction conditions: compound 4c (4.0 mmol), acetone (30 mL), 45 ꢀC, 1h.
Retention time of 4a is the same as that of 4c.
Isolated solid.
ND ¼ not detected.
b
c
d
Please cite this article as: Z. Chen et al., Optimization and process improvement for LCZ696 by employing quality by design (QbD) principles,
Tetrahedron, https://doi.org/10.1016/j.tet.2020.131558

6
Z. Chen et al. / Tetrahedron xxx (xxxx) xxx
Therefore, an operational space was established to minimize the
level of impurity D to less than 0.6% by strictly controlling the
amount of sodium hydroxide and VST within the dark blue area in
Fig. 5.
One advantage from statistically designed experiments is
uncovering optimal conditions and achieving an operational space
that could not be drawn from traditional optimization methodol-
ogies (e.g. OFAT). Hence, taking both product yield and impurity
limit into account, a new control space was generated (white area
in Fig. 6). Parameters indicated by this area was sufficient to gar-
untee a product with acceptable yield (>85.0%) and quality (im-
purity D <0.6%). Under optimized conditions of 1.98 equiv of VST,
5.8 equiv of 46% NaOH aqueous solution and 28 vol of acetone, a
product yield of over 88% was achieved along with a <0.05% con-
tent of impurity D. The results of the three verification experiments
were as expected and were found to be within the 95% CI.
2.5. Recrystallization of LCZ696
Although pure enough to meet specifications, a quality risk
introduced by the hydroscopic property had hindered the acquired
crude product from being final drug substance. To alleviate this
concern, a recrystallization was necessitated. Taking cost and
manipulation suitability into account, we chose acetone which had
been used in the previous step, as the recrystallization solvent after
screening. After dissolution of the main product, the mixture was
filtered to remove excess inorganic salts which were easy to absorb
water from air moisture. However, a normal crystallization could
not facilitate products that satisfy our requirements, neither did a
seed crystallization. To our surprise, after left overnight, solid pre-
cipitation from the solution was observed and the solid was
confirmed non-hydroscopic. Therefore, we further developed an
approach of static crystallization by overnight standing after the
removal of inorganic salts. After the final isolation, the purity of the
final non-hydroscopic drug substance was over 99.9% and all
related impurities were below the specified levels.
Fig. 6. Operating range for NaOH and VST charge.
hydrogenation, N-Boc deprotection and ethyl esterification, and
amidation were allowed to integrate into one robust telescoped
process engendering the only isolated intermediate 4c. Funda-
mental understanding of the process enabled us to maximize
product purity, minimize impurities, facilitate process greenness,
simplify manipulation, shorten production cycle while providing
the same or even better API with acceptable quality. The acquired
calcium sacubitril 4c was subjected to a super-molecular complex
formation with VST under basic conditions and a final purification
to provide LCZ696. The telescoped three steps and the optimization
of the last preparing and purifying step of LCZ696 accounted for a
yield increase by avoiding product lost to the mother liquid during
isolation procedures of intermediates and in seperation of the
product. Finally, the telescoped process was demonstrated at a
150 g scale in the lab furnishing LCZ696 of high purity (>99.9%)
with an overall yield of 73.8% compared with 45.3% of the unte-
lescoped process, reducing production time by 42%.
2.6. Summary of the process development
With various screening techniques and process optimization
under DoE principles, the process was telescoped by avoiding the
isolation of intermediate 2 and 3. The first three steps, asymmetric
2.7. Impurity control strategy
To achieve product with such a high purity and low level of
byproducts, a robost impurity control strategy is required. Typi-
cally, the formation of impurities is determined by the quality of
materials and process parameters such as temperature and equiv-
alents, and the rejection of impurities is usually achieved by
isolation and purification of intermediates. After having evolved
into a telescoped process without isolation and purification of
several intermediates, API needed to be provided in the same
acceptable quality. Through a rigorous understanding of the fate of
the impurities and an operational space established by judicious
evaluation by DoE, the formation of impurities were suppressed
and the impurity rejection was promoted in both intermediate and
API.
Impurity A and B (Fig. 7) are diastereomers of sacubitril. As
mentioned previously, a process impurity in the asymmetric hy-
drogenation step might result in the formation of impurity A.
However, it could also be brought by an epimerization under re-
action conditions (acidic or basic) in the sequential steps. Impurity
B had its root in the enatiomer of starting material 1 [14]. Impurity
C came from a butyl ester derivative impurity in starting material 1
which underwent similar transformations in the downstream
process. Therefore, impurity A, B and C were mainly controlled by
Fig. 5. The contour plot of equivalents of NaOH solution and VST on impurity D.
Please cite this article as: Z. Chen et al., Optimization and process improvement for LCZ696 by employing quality by design (QbD) principles,
Tetrahedron, https://doi.org/10.1016/j.tet.2020.131558

Z. Chen et al. / Tetrahedron xxx (xxxx) xxx
7
By using DoE, a quick detection of optimal conditions was enabled.
We hope that our research will provide an illustration of DoE
application in process development for researchers working in this
field.
4. Experimental section
If no specially indicated, all reagents and solvents were used as
commercially available without further purification. NMR spectra
were measured on a Bruker Avance 400 spectrometer in the solvents
indicated; chemical shifts are reported in units (ppm) by assigning
TMS resonance in the 1H spectrum as 0.00 ppm. Coupling constants
are reported in Hz with multiplicities denoted as singlet (s), doublet
(d), triplet (t), quartet (q), dd (doublet of doublets); multiplets (m),
and etc. HRMS were performed on Fourier Transform Ion Cyclotron
Resonance Mass Spectrometer. Analytical HPLC for liquid phase was
carried out on an Agilent HPLC workstation, equipped with a
CHIRALCEL OJ-RH (4.6 ꢁ 150 mm, 5
mm) column. Gradient elution
hold at 70:30 (0.1% CF3COOH in water)/(acetonitrile) over 16 min,
from 70:30 to 65:35 (0.1% CF3COOH in water)/(acetonitrile) over
5 min, from 65:35 to 60:40 (0.1% CF3COOH in water)/(acetonitrile)
over 19 min, rom 60:40 to 10:90 (0.1% CF3COOH in water)/(aceto-
nitrile) over 10 min, hold for 10 min, then gradient from 10:90 to
70:30 over 0.1 min, hold for 5 min; flow rate 0.7 mL/min; temper-
ature, 25 ꢀC; wave length 254 nm.
4R)-1-([1,10-biphenyl]-4-yl)-5-ethoxy-4-
Fig. 7. Structures of the LCZ696 impurities.
Calcium
(4-(((2S,
methyl-5-oxopentan-2-yl)amino) -4-oxobutanoate) (4c).
(R,E)-5-([1,10-biphenyl]-4-yl)-4-((tert-butoxycarbonyl)amino)-
2-methylpent-2-enoic acid (1) (150 g, 394.2 mmol), LCZ-A (117 mg,
0.12 mmol) and LCZ-B (201 mg, 0.192 mmol) were added to ethanol
(750 mL). The mixture was stirred under 3.0 MPa pressure of
hydrogen gas at 65 ꢀC until starting material 1 was consumed
completely (<2% by HPLC). The resulting mixture containing com-
pound 2 (485.14 g) was used in next step without any further pu-
rification. After an addition of ethanol (1L), SOCl2 (46.4 g,
391.2 mmol) was added slowly to the solution. The temperature
was raised to 38 ꢀC, and the mixture was stirred for 24h. Then the
reaction mixture was concentrated under vacuum at 50ꢂ60 ꢀC to
obtain a crude solid of (2R, 4S)-ethyl 5-([1,10-biphenyl]-4-yl)-4-
amino-2-methylpentanoate hydrochloride (3) (94.37 g).
To a solution of compound 3 in isopropyl acetate (500 mL),
succinic anhydride (35.32 g, 352.95 mmol) and DIPEA (40.67 g,
314.6 mmol) were added at 25 ꢀC and stirred for 4 h. After complete
consumption of starting material (based on HPLC), reaction was
quenched with citric acid (5% aq., 500 mL). The organic layer was
washed with water (500 mL). Brine (1% aq., 1250 mL) was added to
the organic split followed by a slow addition of NaOH (5% aq.,
283 mL) until the pH of the aqueous phase was 7e8. The aqueous
layer was separated and washed with isopropyl acetate (400 mL).
The aqueous split was then evaporated in vacuum at 50 ꢀC until no
significant isopropyl acetate fraction outflows, the temperature was
raised to 95 ꢀC and CaCl2 solution (5% aq, 300 mmol) was added
dropwise. The precipitated solid was filtered and washed with
water (400 mL). The wet solid was dried under vacuum at 75 ꢀC for
24 h to provide 105.5 g (93.5% yield) of the title compound 4c
having 99.21% purity by HPLC. HRMS (m/z) calcd for: 412.2124
tightly restricting impurity levels of starting material 1 and elimi-
natied in the isolation and purification of 4c.
Compared to the above three impurities, the origin of impurity
D was slightly more complicated. In the deprotection and esterifi-
cation step, impurity 7 emerged due to incomplete reaction, which
could further render a formation of impurity D in the following
amidation step. And in the last step of preparing LCZ696, the for-
mation of impurity D was largely impacted by NaOH and VST, ac-
cording to the prior discussion. Design spaces for process
parameters were established applying DoE principles for a mini-
mization of impurity D. As well, all the purification procedures
exhibited a good ability to purge impurity D so that it could be
controlled at less than 0.10% by HPLC in the isolated API.
Impurity E (Fig. 7) is a byproduct emerged from the amidation
step. During the reaction, residue EtOH from the previous depro-
tection and esterification step reacted with excess of succinic an-
hydride and gave this impurity. As the structure of impurity E also
bears a carboxylic group, it could form a salt under basic condition
in the preparation of the final product. Therefore, it is crucial to
strictly limit the residual content of EtOH in crude intermediate 3 to
<3% to ensure an acceptable level of impurity E in both isolated 4c
and API.
3. Conclusion
We have developed a telescoped, viable, and cost effective
process for the manufacture of LCZ696. The reaction process was
optimized under DoE principles, and the impurities of the process
and their control strategies were also investigated. The entire
process was developed with consideration of suitability for scale up
and the manipulation of workup process was simplified to a great
extent compared to known procedures. The synthesis of LCZ696
was achieved by telescoping the process and was successfully
demonstrated at a 150 g scale in the lab providing product within
the quality requirements (purity over 99.9% by HPLC). The overall
yield of the process increased by 28.5% from 45.3% to 73.8%, and the
whole production cycle was also shortened from 12 days to 7 days.
(M4a þ Hþ); found 412.2123；1H NMR (400 MHz, d6-DMSO)
d 7.96
(1H, d, J ¼ 8.2Hz), 7.62 (2H, d, J ¼ 7.4 Hz), 7.55 (2H, d, J ¼ 8.1 Hz), 7.43
(2H, t, J ¼ 7.6 Hz), 7.33 (1H, t, J ¼ 7.3 Hz), 7.24 (2H, d, J ¼ 8.1 Hz),
4.02e3.86 (3H, m), 2.74 (1H, dd, J ¼ 13.4, 6.3 Hz), 2.64 (1H, dd,
J ¼ 13.4, 6.8Hz), 2.50e2.44 (1H, m), 2.26 (4H, dd, J ¼ 22.4,7.3 Hz),
1.81e1.68 (1H, m), 1.47e1.36 (1H, m), 1.10(3H, t, J ¼ 7.1 Hz),1.04 (3H,
s, J ¼ 7.0Hz). 13C NMR (151 MHz, d6-DMSO)
d 179.77, 175.89, 173.65,
172.82, 140.46, 138.55, 138.23, 130.27, 129.34, 127.62, 126.90, 126.77,
60.13, 48.65, 40.98, 40.48, 37.88, 36.49, 33.96, 33.05, 18.37, 14.45.
Please cite this article as: Z. Chen et al., Optimization and process improvement for LCZ696 by employing quality by design (QbD) principles,
Tetrahedron, https://doi.org/10.1016/j.tet.2020.131558

8
Z. Chen et al. / Tetrahedron xxx (xxxx) xxx
N-(3-carboxyl-1-oxopropyl)-(4S)-p-phenylpenylmethyl)-4-
Acknowledgement
amino-2R-methylbut-anoic acid ethylester.[3-((1S, 3R)-1-biphenyl-
4-ylmethyl-3-ethoxycarbonyl-1-butylaminocar-bonyl)pro-
This work was financially supported by the State Key Laboratory
of Anti-Infective Drug Development (Sunshine Lake Pharma
Co.,Ltd), (N0. 2015DQ780357). We also gratefully acknowledge
financial support from the National Natural Science Foundation of
China (21776318).
pionate(S)-30-methyl-2’-(valeryl{2”-( tetrazole -5- yl)biphenyl-40-
ylmethyl}amino) butyric acid] trisodium hemihydrates (LCZ696).
Compound 4c (80 g, 93 mmol) was added into a mixture of iso-
propyl acetate (800 mL) and water (480 mL) followed by an addition
of HCl solution (6% aq., 129.4 g). After complete consumption of the
starting material, the aqueous and organic layers were separated and
washed with citric acid (5% aq., 400 mL ꢁ 2) and water (400 mL ꢁ 2).
The organic layers were concentrated in vacuum at 40ꢂ50 ꢀC to
obtain a colorless viscous liquid. Acetone (1104 mL) was added to the
reaction for a dissolution of the colorless viscous liquid followed by
an addition of VST (80 g, 183.75 mmol). NaOH solution (46% aq.,
539.4 mmol) was added slowly to the resulting mixture, and then
temperature was raised to 45 ꢀC. After having aged for 1 h, the so-
lution was cooled to room temperature and another part of acetone
(1104 mL) was added. A large amount of white solid precipitated out
and was collected by filtration. The white solid was dried under
vacuum at 65 ꢀC for 16 h to afford 154.5 g (86.7% yield) of crude
product of N-(3-carboxyl-1-oxopropyl)-(4S)-p-phenylpenylmethyl)
-4-amino-2R-methylbut-anoic acid ethylester. [3-((1S, 3R)-1-
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.tet.2020.131558.
References
[1] L. Feng, P.H. Karpinski, P. Sutton, Y. Liu, D.F. Hook, B. Hu, T.J. Blacklock,
P.E. Fanwick, M. Prashad, S. Godtfredsen, C. Ziltener, Tetrahedron Lett. 53
(2012) 275e276.
[2] According to U.S. Food and drug administration (FDA), last acess date, https://
www.accessdata.fda.gov/scripts/cder/daf/index.cfm?event¼BasicSearch.
process. (Accessed 23 July 2020).
[3] (a) J. Yuan, Y. Huang, Y. Song, CN Patent 105985225 (A) (2015);
(b) M. Liu, Y. Wang, J. Wu, CN Patent 105168205 (A) (2015);
(c) S. Li, W. Xu, H. Hua, CN Patent 106397248 (A) (2017).
[4] Bollu R.B., Thaimattam R., Challagonda V.R., Yasam S.R., Seethamraju P.K.,
Vasireddi U.M.R., Patent Appl. WO 2018/069833 A1 (2018).(b) Q. Zhang,
Q. Chen, L. Feng, Y. He, CN Patent 105622535 (A) (2016).
[5] Feng L., Godtfredsen S.E., Karpinski P., Sutton P.A., Prashad M., Girgis M.J., Hu
B., Liu Y., Pharmaceutical combinations of an angiotensin receptor antagonist
and an nep inhibitor, Patent Appl. WO 2017/009784 A1 (2007).(b) K.K. Singh,
S.D. Diwakar, S.D. Jadhav, Patent Appl. WO (2017).
[6] Prasad V.S.R., Sahu A., Shimpi N.A., Patent Appl. WO 2016/135751 A1 (2016).
[7] Y. Wang, Z. Qiao, Z. Li, CN Patent 108373423 (A) (2018).
[8] Zhu G., Tu Y., Xie W., Wang Z., Xu S., Patent Appl. WO 2017/096772 A1 (2017).
[9] X. Gu, J. Zhao, L. Chen, Y. Li, B. Yu, X. Tian, Z. Min, S. Xu, H. Gu, J. Sun, X. Lu,
M. Chang, X. Wang, L. Zhao, S. Ye, H. Yang, Y. Tian, F. Gao, Y. Gai, G. Jia, J. Wu,
Y. Wang, J. Zhang, X. Zhang, W. Liu, X. Gu, X. Luo, H. Dong, H. Wang,
B. Schenkel, F. Venturoni, P. Filipponi, B. Guelat, T. Allmendinger, B. Wietfeld,
P. Hoehn, N. Kovacic, L. Hermann, T. Schlama, T. Ruch, N. Derrien, P. Piechon,
F. Kleinbeck, J. Org. Chem. 85 (2020) 6844e6853.
biphenyl-4-ylmethyl-3-ethoxycarbonyl-1-butylaminocar-
bonyl)
propionate-(S)-30-methyl-2’-(valeryl{2”-( tetrazole -5- yl)biphenyl-
40-ylmethyl}amino) butyric acid] trisodium hemihydrates (5) with
99.83% purity by HPLC. Crude product 5 (100 g, 17.4 mmol) was
added into a reactor charged with acetone (1500 mL) the mixture
was heated to reflux. After the majority of the product was dissolved,
a filtration was performed to remove residual inorganic salts. Then
the filtrate was cooled to room temperature slowly and allowed for
an overnight static crystallization. The white solid precipitated out
was collected by filtration and dried under vacuum at 65 ꢀC for 24 h
to afford 91.01 g (91% yield from 5) of LCZ696 with 99.98% purity by
HPLC. HRMS (m/z) calcd for 412.2124 (M4a þ Hþ) and 436.2349
(MVST þ Hþ); found 412.2112 and 436.2336.1H NMR (400 MHz, d6-
DMSO)
d
8.05 (1H, d, J ¼ 8.4 Hz), 7.63 (2H, d, J ¼ 7.6 Hz), 7.56 (2H,
d, J ¼ 8.0 Hz), 7.51e7.48 (1H, m), 7.43 (2H, t, J ¼ 7.6 Hz), 7.40e7.36
(1H, m), 7.35e7.27 (3H, m), 7.24 (2H, d, J ¼ 8.4 Hz), 7.04 (2H, d,
J ¼ 8.4 Hz), 6.92 (2H, d, J ¼ 8.0 Hz), 4.58 (1H, dd, J ¼ 23.2, 10.4 Hz),
4.47e4.37 (1H, m), 3.99e3.94 (2H, m), 3.92e3.83 (2H, m), 3.79e3.78
(1H, m), 3.66 (2H, s), 2.71 (1H, dd, J ¼ 13.2, 6.4 Hz), 2.61 (1H, dd,
J ¼ 13.6, 6.8 Hz), 2.45e2.33 (1H, m), 2.20 (2H, t, J ¼ 7.0 Hz), 2.11 (2H, t,
J ¼ 7.2 Hz), 2.05e1.93 (1H, m), 1.77e1.65 (1H, m), 1.56e1.46 (1H, m),
1.41e1.33(1H, m), 1.28 (2H, dd, J ¼ 14.4, 6.8 Hz), 1.09 (3H, t,
J ¼ 7.0 Hz), 1.02 (3H, d, J ¼ 7.2 Hz),0.87 (3H, d, J ¼ 6.0 Hz), 0.85 (2H, d,
J ¼ 7.6 Hz), 0.69 (2H, t, J ¼ 7.4 Hz), 0.60 (2H, d, J ¼ 6.8 Hz);13C NMR
[10] ICH’s Q8 and Q11 Guidelines Have Identified DoE as the Key Technology
Module of QbD and Applied it to the Development of Pharmaceutical Tech-
nologies (API Process Development, Analytical Method Development, Prepa-
ration Process Development)
[11] (a) According to U.S. Food and Drug Administration (FDA), MAPP 5016.1
Applying ICH Q8(R2) Q9 and Q10 Principles to CMC Review (b) According to
European Medicines agency (EMA). https://www.ema.europa.eu/en/human-
regulatory/research-development/quality-design.
[12] (a) S.A. Weissman, N.G. Anderson, Org. Process Res. Dev. 19 (2015)
1605e1633;
(b) DoE, Design of Experiment) Is a Mathematical Statistics Method to Arrange
Experiments and Analyze Experimental Data. Through the Reasonable
Arrangement of Experiments, the Ideal Experimental Results and Scientific
Experimental Conclusions Can Be Obtained with a Smaller Experiment Scale,
Fewer Experiment Times, Shorter Experiment Period and Lower Experiment
Cost. Min, Y. Application Guide of Design of Experiments, DOE), China Machine
Press, 2010.
(151 MHz, d6-DMSO)
d 176.11, 174.42, 172.40, 171.36, 170.70, 155.52,
141.61, 140.39, 138.44, 138.15, 138.08, 131.67, 130.99, 130.27, 129.42,
129.24, 128.75, 127.70, 127.37, 126.94,126.53, 123.70, 67.68, 66.32,
63.38, 60.28, 49.07, 48.75, 40.89, 37.86, 36.38, 32.93, 30.65, 29.71,
27.93, 27.42, 27.22, 22.31, 21.88, 21.51, 20.48, 19.79, 19.13, 18.86, 18.31,
14.37, 14.09.
[13] (a) K. Gary, Patent Appl. US 5217996 (A) (1993);
(b) D. Hook, B. Wietfeld, M. Lotz, Patent Appl. WO 2008/031567 (2008);
(c) Liu, Q.; Sun, J.; Chen, H.; Song, G.; Cong, R.; Shen, W. CN106318988 (A),
2017. (d) X. Zhang, T. Zhou, Z. Xiong, H. Zhou, T. Hu, CN Patent108238981 (A)
(2018);
Declaration of competing interest
(e) X. Pan, Y. Ge, T. Xia, CN Patent 106631903 (A) (2017);
(f) S. Wu, H. Jiang, H. Wang, C. Jiang, J. Yang, Q. Xu, CN Patent 108084058 (A)
(2018);
(g) X. Liu, Y. Dai, Z. Zhang, F. Yu, W. Huang, CN Patent 105061263 (A) (2015);
(h) X. Zhang, CN Patent 105481622 (A) (2016).
The authors declare that they have no known competing
financial interests or personal relationships that could have
appeared to influence the work reported in this paper.
[14] A. Halama, M. Zapadlo, Org. Process Res. Dev. 23 (2019) 102e107.
Please cite this article as: Z. Chen et al., Optimization and process improvement for LCZ696 by employing quality by design (QbD) principles,
Tetrahedron, https://doi.org/10.1016/j.tet.2020.131558