App oaching d ug elease pe o mance om mesopo ous silica
o mula ions by modeling o chemical po en ials
And eas Niede quell
a,b
, Annika Ho e
b
, Ba bo a V aníko ´
a
a
, Ma in Kuen z
b,*
a
Depa men o Pha maceu ical Technology, Facul y o Pha macy in H adec K ´
alo ´
e, Cha les Uni e si y, Akademika Hey o sk´
eho 1203, 500 05 H adec K ´
alo ´
e, Czech
Republic
b
Ins i u e o Pha ma Technology, Uni e si y o Applied Sciences and A s No hwes e n Swi ze land, School o Li e Sciences FHNW, Ho acke s . 30, 4132 Mu enz,
Swi ze land
ARTICLE INFO
Keywo ds:
D ug dissolu ion
Chemical po en ials
Silica-wa e pa i ioning coe icien
Non-o de ed mesopo ous silica
COSMO-RS
D ug-silica in e ac ions
ABSTRACT
Mesopo ous silica a e p omising bio-enabling ca ie s o poo ly soluble d ugs. Howe e , a comp ehensi e un-
de s anding o d ug-silica in e ac ions and hei impac on d ug elease emains limi ed. Apa om u gen ly
needed expe imen al ools, p edic i e in silico ools ha conside d ug-ca ie in e ac ions in aqueous media a e
cu en ly lacking. To add ess his gap, a no el in silico app oach (silica-wa e pa i ioning coe icien ) was
in oduced in his s udy. A se ies o en d ugs we e loaded on o a mesopo ous ca ie (Pa eck® SLC 500), and
he p oduc s we e analyzed using di e en ial scanning calo ime y (DSC) and X- ay powde di ac ion (XRPD).
In i o dissolu ion (USP II) p o iles o d ug-loaded o mula ions we e analyzed and co ela ed wi h a newly
in oduced silica-wa e pa i ioning coe icien de i ed om chemical po en ial calcula ions using he
Conduc o -like Sc eening Model o Real Sol en s (COSMO-RS). S ong co ela ions we e obse ed be ween
dissolu ion pa ame e s, such as he ini ial elease slopes (Pea son =-0.98; p =<0.05) and AUC alues (Pea son
=-0.79; p <0.05), and he calcula ed chemical po en ial-based pa i ioning coe icien . This s udy in oduces a
p edic i e me hod based on COSMO-RS-de i ed chemical po en ials o es ima e silica-wa e pa i ioning o
d ugs, he eby p edic ing hei elease pe o mance om mesopo ous silica o mula ions. The esul s demon-
s a e ha hese calcula ed chemical po en ials can quali a i ely ank he d ug elease kine ics in aqueous media.
Fu he in es iga ion wi h addi ional compounds and ca ie ypes may b oaden he applicabili y o his
app oach as a mechanis ic ool o mesopo ous silica o mula ion de elopmen and con ibu e o na owing he
gap owa d u u e clinical ansla ion.
1. In oduc ion
Mesopo ous silica-based o mula ions (MS) and amo phous solid
dispe sions (ASDs) a e well-es ablished o mula ion s a egies o
enhance he solubili y and s abili y o poo ly wa e -soluble d ugs by
s abilizing he d ug in a non-c ys alline o m (A gyo e al., 2014;
B emmell and P es idge, 2019; Kawaba a e al., 2011; Khan e al., 2022;
Maleki e al., 2017; McCa hy e al., 2016; Mehmood e al., 2017; Na -
ayan e al., 2018; P es idge e al., 2007; Qian and Bogne , 2012; Rogge s
e al., 2014; San os e al., 2011; Tang e al., 2012; Valle -Regí e al.,
2018). ASDs ypically consis o a d ug molecula ly dispe sed in a
polyme ma ix ha s abilizes he amo phous o m by inc easing he
glass ansi ion empe a u e and educing molecula mobili y.
Supe sa u a ion upon dissolu ion is o en main ained h ough
polyme -d ug in e ac ions and he inhibi ion o nuclea ion and c ys al
g ow h (Chiou and Keyph ases, 1971; Leune and D essman, 2000). In
con as o polyme -based amo phous solid dispe sions, mesopo ous
silica (MS) o mula ions u ilize ino ganic ca ie s wi h de ined meso-
po es o con ine d ug molecules in a spa ially es ic ed en i onmen .
Acco ding o he IUPAC classi ica ion, mesopo ous ma e ials possess
po e diame e s be ween 2 and 50 nm, a ange pa icula ly ele an o
d ug deli e y, as i enables e icien adso p ion while pe mi ing mo-
lecula mobili y o deso p ion. This physical con inemen , in combi-
na ion wi h in e acial in e ac ions a he silica su ace, can supp ess
c ys alliza ion and modula e d ug elease kine ics. Al hough such sys-
ems o en ha e limi ed d ug-loading capaci y and ypically do no
* Co esponding au ho a : Uni e si y o Applied Sciences and A s No hwes e n Swi ze land, Pha ma Technology and Bio echnology, Ho acke s . 30, CH-4132
Mu enz Swi ze land.
E-mail add ess: [email p o ec ed] (M. Kuen z).
Con en s lis s a ailable a ScienceDi ec
Eu opean Jou nal o Pha maceu ical Sciences
jou nal homepage: www.else ie .com/loca e/ejps
h ps://doi.o g/10.1016/j.ejps.2025.107283
Recei ed 15 July 2025; Recei ed in e ised o m 10 Sep embe 2025; Accep ed 18 Sep embe 2025
Eu opean Jou nal o Pha maceu ical Sciences 214 (2025) 107283
A ailable online 19 Sep embe 2025
0928-0987/© 2025 The Au ho s. Published by Else ie B.V. This is an open access a icle unde he CC BY license ( h p://c ea i ecommons.o g/licenses/by/4.0/ ).
inhe en ly main ain supe sa u a ion in aqueous media, hey can o e
ad an ages o e con en ional ASD app oaches based on ho -mel
ex usion o sp ay d ying. These ad an ages may include imp o ed
physical s abili y h ough spa ial con inemen , as well as enhanced
chemical s abili y (Di zinge e al., 2019; Zhang e al., 2022). The ma-
e ial used in his s udy, Pa eck® SLC 500, is a diso de ed mesopo ous
silica wi h a high speci ic su ace a ea and an a e age po e diame e
wi hin he mesopo ous ange (6.1 ±0.01 nm) (V aníko ´
a e al., 2020).
I has been p e iously applied in s udies on d ug-silica in e ac ions and
loading-dependen amo phous s abili y (Niede quell e al., 2023;
V aníko ´
a e al., 2020). Beyond hei s abilizing capaci y, MS ma e ials
a e pha maceu ically a ac i e due o hei GRAS (Gene ally Recog-
nized as Sa e) s a us and abili y o shield d ugs om mois u e, oxygen,
and ligh .
In he li e a u e, MS ha e equen ly been discussed in he con ex o
s abilizing poo glass o me s i.e., compounds wi h low glass- o ming
abili y (GFA class I and II) ha a e p one o ec ys alliza ion and o en
equi e p ecipi a ion inhibi o s o supe sa u a ion main enance (Bai d
e al., 2010; Wy enbach e al., 2016). Howe e , MS d ug deli e y is
equally ele an o GFA class III compounds, which a e in insically
mo e s able in hei amo phous o m and end o emain me as able in
supe sa u a ed solu ions, o en wi hou equi ing polyme ic p ecipi a-
ion inhibi o s. Al hough MS-based o mula ions ha e ye o gain
widesp ead adop ion in he pha maceu ical ma ke , hei unique capa-
bili ies in s abilizing and deli e ing poo ly soluble d ugs a e d i ing
ongoing esea ch and de elopmen e o s, making hem a e sa ile d ug
deli e y pla o m. Howe e , a signi ican knowledge gap emains
ega ding he molecula in e ac ions be ween indi idual d ugs and silica
su aces and how hese in e ac ions in luence s abili y and d ug elease.
Besides his esea ch gap, mos publica ions ocus on only one o h ee
model compounds, e en hough i is impo an o s udy a b oade
da ase . Addi ionally, he pe o mance o MS o mula ions is o en
benchma ked agains c ys alline d ug o ms using dissolu ion expe i-
men s, which a e ime-consuming. The e o e, i would be bene icial o
ha e ei he compu a ional o simple expe imen al ools o an icipa e
elease pe o mance be o e emba king on ime-consuming biopha ma-
ceu ical in i o expe imen a ion. To app oach hese esea ch gaps, ou
ecen in es iga ions explo e in insic and appa en molecula in-
e ac ions be ween d ugs and mesopo ous silica, pa icula ly conce ning
o mula ion s abili y (Niede quell e al., 2025, 2023). While in insic
a ini y ep esen s a di ec bina y in e ac ion be ween he d ug and
silica su ace, an appa en in e ac ion occu s in he p esence o a sol en
o wa e phase.
An impo an expe imen al me hodology in his con ex has been he
use o so-called in si u so p ion s udies, as demons a ed by he g oup o
Lynne Tailo , which p o ide aluable insigh in o d ug adso p ion on o
mesopo ous silica su aces (Dening e al., 2019; McCa hy e al., 2020).
The d ug loading so p ion me hod in ol es he adso p ion o d ug
molecules om a solu ion on o mesopo ous silica ma e ials based on
in e ac ions be ween he d ug and he silica su ace. When he d ug is in
a supe sa u a ed solu ion, i can adso b on o he la ge su ace a ea o
silica, o ming bo h monolaye s and mul ilaye s wi hin he mesopo es.
The d i ing o ces o his so p ion include hyd ogen bonding be ween
he d ug and su ace silanol g oups o he silica, and o he in e ac ion
mechanisms like elec os a ic o ces and Van de Waals o ces (Eugene
Papi e , 2000; Ha e e al., 2020a; Niede quell e al., 2025).
Apa om expe imen al s udies o d ug-ca ie in e ac ions, much
wo k has been done also in silico, bu mos app oaches ocused on he
bina y in e ac ion o d ug and silica (Delle Piane e al., 2014; Gignone
e al., 2015; Na ayan e al., 2022; Niede quell e al., 2023). Acco dingly,
hese molecula modeling app oaches calcula e he in insic in e ac ions
be ween he d ug and he mesopo ous silica ca ie , which may be use ul
o d ug loading o s abili y conside a ions. Howe e , hei ele ance o
d ug elease emains unclea because wa e has no been explici ly
conside ed. Wa e has been o he wise conside ed in, o example, o he
p edic ions based on he conduc o -like sc eening model (COSMO),
which has been used in he pha maceu ical ield, anging om solubili y
p edic ions (Klajmon, 2022; Klam e al., 2002; Mac Fhionnlaoich e al.,
2024) o pa i ioning analyses (Mahmoudabadi and Pazuki, 2020;
Wa nau e al., 2021). These p io applica ions highligh he abili y o he
model o so-called “ eal sol en s”, i.e., COSMO-RS, o cap u e sol a ion
and pa i ioning phenomena ha ha e pha maceu ical ele ance.
The e o e, i would be in e es ing o use such modeling also in he
con ex o d ug-silica in e ac ions in an aqueous en i onmen .
A pa icula s eng h o COSMO-RS is he combina ion o quan um-
chemical su ace cha ge p edic ions wi h he modynamic modeling
(Loschen and Klam , 2015). By calcula ing he chemical po en ials o he
d ug in a silica o wa e en i onmen , a silica-wa e pa i ioning coe -
icien can be ob ained. We es ed whe he a COSMO-RS-de i ed sili-
ca-wa e pa i ion coe icien could be compa ed and co ela ed wi h
dissolu ion kine ics o en mesopo ous silica o mula ions. Ra he han
aiming o op imize o mula ion pa ame e s such as supe sa u a ion o
physical s abili y, he ocus was on assessing whe he he calcula ed
a ini ies owa d silica and wa e can explain he ex en and a e o d ug
deso p ion in o aqueous media. To his end, d ug loading was es ic ed
o he heo e ical monolaye adso p ion capaci y, he eby minimizing
po en ial in e e ence om c ys alliza ion o in e molecula d ug-d ug
in e ac ions. In addi ion, p ima ily glass- o ming abili y (GFA) class III
compounds we e selec ed o hei good amo phous s abili y, educing
he likelihood o d ug c ys alliza ion ei he du ing loading o on d ug
elease, which would bo h be a pe u bing ac o in he amewo k o he
cu en s udy o compa e he silica pa i ion coe icien wi h expe i-
men al elease pe o mance. As he la e quali y a ibu es o d ug
elease, he loga i hmic ini ial slope o d ug elease a e and he loga-
i hmic a ea unde he concen a ion- ime cu e (AUC) we e consid-
e ed. Beyond iden i ying co ela ions, his s udy also sough o p o ide
mechanis ic insigh s in o d ug elease beha io om mesopo ous ca -
ie s, suppo ing ea ly p e o mula ion wi h he assessmen o such a
o mula ion s a egy and la e guiding he a ional de elopmen o
silica-based d ug p oduc s.
2. Ma e ials and me hods
2.1. Ma e ials
The model d ugs (Fig. 1), including ca edilol, diclo enac ( ee acid),
i e mec in, ke oconazole, lopina i , pimozide, p obucol, and i ona i ,
we e ob ained om Biosyn h L d. (Comp on, Uni ed Kingdom), and
indome hacin was pu chased om Ac os O ganics L d. (Geel, Belgium),
and quinidine was ecei ed om Sigma-Ald ich Chemie L d. (Buchs,
Swi ze land). The diso de ed mesopo ous silica ca ie Pa eck® SLC
500 was kindly supplied by Me ck KGaA (Da ms ad , Ge many).
O ganic sol en s such as ace one, ace oni ile, dichlo ome hane, and
me hanol, as well as o he chemicals such as po assium dihyd ogen
phospha e, sodium hyd oxide, and dioc yl sul osuccina e sodium sal
(DOSS), we e ob ained om Me ck KGaA. Ammonium ace a e and
ace ic acid we e pu chased om Fluka Analy ical - Honeywell Inc.
(Cha lo e, USA). O ho-phospho ic acid was p ocu ed om Ca l Ro h
GmbH & Co. KG (Ka ls uhe, Ge many). The wa e used h oughou he
s udy was p epa ed wi h an A ium® 61,215 wa e pu i ica ion sys em
om Sa o ius S edim Bio ech L d. (G¨
o ingen, Ge many). All excipien s
we e used as supplied, wi hou u he pu i ica ion.
2.2. Me hods
2.2.1. D ug s a e cha ac e iza ion
2.2.1.1. Di e en ial scanning calo ime y (DSC). Di e en ial scanning
calo ime y was used o cha ac e ize he physical s a e o he d ug
subs ances. The measu emen s we e ca ied ou wi h he STARe Sys em
DSC 3 om Me le Toledo GmbH (G ei ensee, Swi ze land). The hea
A. Niede quell e al.
Eu opean Jou nal o Pha maceu ical Sciences 214 (2025) 107283
2
low di e ence be ween he wo independen u naces was measu ed,
wi h he blank u nace used as a e e ence. The pu e ac i e ing edien s,
blank mesopo ous silica ca ie , and d ug-loaded silica powde s we e
measu ed. In addi ion, physical mix u es consis ing o silica pa icles
and ac i e ing edien s we e analyzed a a a io co esponding o he
monolaye loading capaci y (MLC). Fo his pu pose, 5 o 10 mg o he
sample was placed and c imped in a s anda d 40
μ
L aluminum pan wi h
pie ced lids. The samples we e subjec ed o wo hea ing and wo cooling
cycles. The empe a u e was a ied om 25 o 270 ◦C wi h a hea ing/
cooling a e o 10 ◦C/min. The da a we e analyzed using he STARe
so wa e.
The glass o ming abili y (GFA) o he d ug subs ances was classi ied
based on hei c ys alliza ion beha io du ing he mal p ocessing using
he same DSC measu emen me hod. A mo e de ailed desc ip ion o he
me hod can be ound in he li e a u e (Bai d e al., 2010). Compounds
we e ca ego ized in o h ee GFA classes acco ding o es ablished
c i e ia:
GFA Class I (poo glass o me s): These compounds c ys allize
eadily upon cooling om he mel , indica ing a low esis ance o
ec ys alliza ion and poo glass s abili y.
GFA Class II (in e media e glass o me s): These subs ances o m an
amo phous glass upon cooling bu end o ec ys allize ei he du ing
s o age o upon subsequen ehea ing, sugges ing mode a e kine ic
esis ance o c ys alliza ion.
GFA Class III (good glass o me s): These compounds emain amo -
phous upon bo h cooling and ehea ing, demons a ing high glass
s abili y and a minimal c ys alliza ion endency om he mel .
2.2.1.2. X- ay powde di ac ion (XRPD). In addi ion o he measu e-
men s o pu e d ugs, X- ay di ac og ams we e eco ded o con i m he
amo phous cha ac e and e alua e he s abili y o he physical mix u es
and d ug-loaded mesopo ous silica o mula ions. The e o e, an X- ay
di ac ome e (R-XRD Phase D2) om B uke AXS L d. (Ka ls uhe,
Ge many) was employed. The de ice was equipped wi h a Co and Cu
KFL ube (30 kV, 10 mA) as he adia ion sou ce and a 1-D Lynxeye®
de ec o . The measu emen s we e ca ied ou a a ol age o 30 kV and a
cu en o 10 mA. The sample was o a ed a 15 pm, and he angua
scanning ange o each sample was om 6◦(2θ) o 40◦(2θ) wi h a s ep
size o 0.02◦(2θ) a 2.0 s pe s ep. Unde hese condi ions, he de ec ion
limi o he c ys alline con en was app oxima ely 1 % w/w.
2.2.2. Loading o he mesopo ous silica pa icles
The monolaye capaci y (MLC), which is he d ug loading o a
heo e ical monolaye adso p ion on mesopo ous silica su aces, was
calcula ed using Eq. (1). This app oach has been well desc ibed in he
Fig. 1. 2D chemical s uc u es o 10 selec ed pha maceu ical ac i e subs ances.
A. Niede quell e al.
Eu opean Jou nal o Pha maceu ical Sciences 214 (2025) 107283
3
li e a u e (Ba nhøj e al., 2019; Dening and Taylo , 2018; Le e al.,
2019). The speci ic su ace a ea (SSA) o he u ilized silica pa icles was
aken om p e iously published measu emen s (Niede quell e al.,
2023) using ni ogen and he “B unaue Emme Telle ” (BET) me hod.
N
A
ep esen s A ogad o’s numbe , M
w
is he molecula weigh , and o
he maximum p ojec ed con ac a ea (SA
M
), he wo la ges molecula
dimensions o he indi idual d ug molecules we e used, which we e
es ima ed using he Molecula Modeling P o Plus 8.2.1 so wa e
(No gwyn Mon gome y So wa e Inc., No h Wales, USA).
MLC [%,g/g] = SSA⋅Mw⋅1020
SAM⋅NA
(1)
The d ug-loading o he mesopo ous silica powde (Pa eck® SLC
500) was ca ied ou acco ding o he incipien we ness imp egna ion
me hod (Khalbas e al., 2024a, 2024b). Fo he loading p ocess, a
Chemyx Fusion 200 sy inge pump om Chemyx Inc. (S a o d, USA) was
employed. Be o e loading, he silica pa icles we e d ied o 12 h a 100
◦C and a acuum o 550 mba in a acuum d ying o en o ype KVST-11
om Sal is Plc. (Reussbühl, Swi ze land). Two di e en sol en s we e
used o loading due o he di e en solubili ies o he ac i e ing edien s.
Ace one was used o ca edilol, diclo enac ( ee acid), and lopina i ,
while dichlo ome hane was he choice o indome hacin, i e mec in,
ke oconazole, pimozide, p obucol, quinidine, and i ona i . All d ug
solu ions had a concen a ion o 20 mg/mL and we e added d opwise o
1 g o mesopo ous silica a a a e o 0.1 mL/min while using con inuous
mild s i ing (50 pm) o he mesopo ous silica (MS) ca ie on a mag-
ne ic s i ing pla e. Using a p ecise au oma ed sy inge pump o he
d opwise d ug loading, a ypical accu acy o abou ±2 % can be ach-
ie ed. Subsequen ly, he samples we e placed in a d ying cabine o
allow sol en e apo a ion. Taking he calcula ed MLC in o accoun (see
Table 1), he MS ca ie was loaded wi h di e en olumes o d ug so-
lu ion. The loaded silica samples we e hen le o 24 h a 60 ◦C and a
acuum o 500 mba in a KVTS-11 d ying o en. This empe a u e was
chosen due o he boiling poin s o he sol en s: ace one ~ 56 ◦C and
dichlo ome hane ~ 40 ◦C. The loaded mesopo ous silica samples we e
hen sealed ai igh and s o ed in a desicca o o p o ec hem om
mois u e un il u he use. The ac ual gained mass a e d ug-loading
and d ying was e i ied g a ime ically and kep as close as possible
o he calcula ed MLC. The e o e, he silica pa icles we e weighed on
he XS205DU analy ical balance om Me le Toledo GmbH (G ei ensee,
Swi ze land) wi h a eadabili y o 0.01 mg. The mass gain in pe cen ( %
g/g) was de e mined based on he weigh di e ence be ween he d ied
emp y silica ca ie and ha a e d ug loading and d ying.
2.2.3. Solubili y de e mina ion
Samples (n =3) we e p epa ed by adding excess d ug o 4 mL o
dissolu ion medium (phospha e bu e a pH 6.8 wi h 0.5 % (w/ )
dioc yl sul osuccina e sodium sal (DOSS)). The samples we e le a
oom empe a u e (25 ◦C) o 24 h wi h con inuous s i ing on a mag-
ne ic s i e a 200 pm. Subsequen ly, o sepa a e undissol ed d ug
pa icles, he samples we e cen i uged a 15,000 pm o 10 min in an
Eppendo 5420 mic ocen i uge om Eppendo Plc. (Hambu g, Ge -
many) and addi ionally il e ed wi h i anium HPLC sy inge il e s
(diame e : 17 mm, memb ane: PTFE, po e size: 0.45
μ
m). The samples
we e dilu ed wi h dissolu ion medium, and he concen a ions we e
de e mined using HPLC.
2.2.4. Dissolu ion measu emen s
Fo he elease o he loaded silica o mula ions, a USP ype II (model
DT 600) dissolu ion appa a us om ERWEKA, Plc. (Langen, Ge many)
was used. A phospha e bu e a pH 6.80 ±0.02 wi h 0.5 % (w/ ) DOSS
was used as he elease medium. The pH was adjus ed o 6.8 wi h 1 M
sodium hyd oxide solu ion be o e adding 0.5 % (w/ ) DOSS. This su -
ac an concen a ion was selec ed o ensu e su icien we ing and
dispe sion o he mesopo ous silica while a oiding no able micella
solubiliza ion, which could obscu e d ug-silica in e ac ion e ec s. The
paddle posi ion was adjus ed o 25 ±2 mm abo e he essel bo om in
acco dance wi h USP equi emen s. Each essel con ained 250 mL o
elease medium, and he paddle speed was se a 25 pm. In each essel,
120 mg o he d ug-loaded mesopo ous silica o mula ion was in o-
duced, whe e he d ug amoun a ge ed he MLC o he gi en d ug. A
in e als o 2, 3, 4, 5, 10, 15, 20, 30, 40, 50, 60, 75, 90, 120, 150, and
180 min, 2 mL each we e sampled and eplaced wi h he same olume o
esh phospha e bu e . The collec ed samples we e il e ed using an
HPLC i anium sy inge il e (diame e : 17 mm and 0.45
μ
m po e size o
he PTFE memb ane) in o HPLC ials. The concen a ion o he samples
was hen analyzed ia HPLC in iplica es.
2.2.5. Concen a ion de e mina ion using high-pe o mance liquid
ch oma og aphy (HPLC)
The HPLC measu emen s we e pe o med on an Agilen In ini y 1260
II sys em (Agilen Technologies GmbH, Waldb onn, Ge many), which
was equipped wi h he ollowing componen s: G7167A mul isample ,
G7112B bina y pump, G1316A column o en (Agilen 1100 se ies), and
G1314B a iable wa eleng h de ec o (Agilen 1200 se ies). The column
used was an XB idge BEH C18 om Wa e s L d. (Eschbo n, Ge many)
wi h a po e size o 130 Å, pa icle size o 5
μ
m, inne diame e o 4.6 mm,
and leng h o 150 mm. Da a acquisi ion and e alua ion we e done wi h
he Agilen OpenLab CDS ChemS a ion so wa e Re . C.01.10 [287].
Va ious mobile phases we e used o HPLC analysis. Ace oni ile was
used in combina ion wi h ei he pu i ied wa e (B1), pu i ied wa e wi h
0.1 % phospho ic acid ( / ) a a pH o 2.3 (B2), o pu i ied wa e wi h
0.1 % ace ic acid ( / ) and 10 mM ammonium ace a e a a pH alue o
4.5 (B3). P io o use, he mobile phases we e addi ionally il e ed
h ough a polye he sul one (PES) memb ane il e (po e size: 0.22
μ
m,
diame e : 47 mm) om Memb ane Solu ions, LLC. (Aubu n, USA) a a
acuum o 300 mba . Calib a ion cu es we e made by p epa ing s ock
solu ions wi h 1 mg/mL o he indi idual d ugs in ace oni ile. The s ock
solu ions we e il e ed h ough 17 mm-sized i anium HPLC sy inge
il e s (memb ane: PTFE, po e size: 0.45
μ
m) om In och oma Plc.
(Goldau, Swi ze land). The heo e ical monolaye capaci y (MLC) was
conside ed, and he s ock solu ions we e u he dilu ed o 1:2, 1:4, 1:8,
1:16, 1:32, 1:64, and 1:100 ( / ). Each dilu ion was p epa ed and
measu ed in iplica es. The measu emen me hods we e op imized
based on p e-expe imen al uns, and o he concen a ion de-
e mina ions o each API, indi idual se ings, such as mobile phase
composi ions and wa eleng hs, we e chosen (see Table 1). The mea-
su emen s we e pe o med a 25 ◦C, an injec ion olume o 10
μ
L, and a
low a e o 0.8 mL/min, excep o lopina i , whe e a low a e o 0.6
mL/min was used, we e employed.
2.2.6. Molecula modeling - calcula ion o chemical (pseudo-) po en ials
Chemical po en ials o d ug in a silica phase (
μ
s
(silica)
) o aqueous
en i onmen (
μ
s
(wa e )
) and he chemical po en ial o sel -associa ion o
Table 1
Calcula ed heo e ical Monolaye Capaci y (MLC), mobile phase composi ion,
and UV wa eleng h o he HPLC concen a ion de e mina ion o 10 indi idual
d ugs. Eluen s A and Bx a e speci ied in he ex .
Subs ance name MLC [% g/g] Mobile phase
composi ion eluen
A [% / ] and gi en
Bx [% / ]
Wa eleng h [nm]
Ca edilol 16.1 60 40 (B2) 240
Diclo enac ( ee acid) 22.1 70 30 (B3) 270
Indome hacin 18.6 70 30 (B3) 260
I e mec in 22.3 95 5 (B1) 245
Ke oconazole 12.4 70 30 (B3) 244
Lopina i 19.3 70 30 (B1) 215
Pimozide 19.5 70 30 (B3) 280
P obucol 22.6 98 2 (B1) 242
Quinidine 21.2 70 30 (B3) 235
Ri ona i 19.9 70 30 (B3) 215
A. Niede quell e al.
Eu opean Jou nal o Pha maceu ical Sciences 214 (2025) 107283
4
d ug molecules (µs
(sel )
) we e calcula ed using COSMOquick . 2020
om BIOVIA, Dassaul sys ems SE (V´
elizy-Villacoublay Cedex, F ance).
10 d ugs we e assigned as solu es, whe eas a wa e molecule and a small
slab o amo phous silica we e used as he molecula basis o modeling
o he phases o d ug pa i ioning. Fig. 2 depic s a schema ic ep e-
sen a ion o he en i e COSMO-RS wo k low o ob aining chemical
(pseudo-) po en ials. In b ie , COSMO-RS is based on Densi y Func ional
Theo y (DFT) calcula ions o gene a e a disc e e molecula su ace
embedded in a i ual conduc o (Klam , 2011). Each segmen o his
su ace is cha ac e ized by i s a ea and shielding cha ge densi y (
σ
),
which e lec s he elec os a ic sc eening by he en i onmen and he
back-pola iza ion o he solu e molecule. The esul ing su ace cha ge
densi y dis ibu ion is hen adap ed o he ac ual dielec ic cons an o
he sol en , e ec i ely simula ing he elec os a ic in e ac ions be ween
he molecule and he sol en (Ecke and Klam , 2002). While he su -
ace cha ge densi y dis ibu ion is ob ained by DFT calcula ions, sub-
sequen s a is ical he modynamics o hese su ace segmen s a e used o
de e mine he chemical po en ials (
μ
) o molecules in liquids (Klam ,
2005; Klam e al., 1998). Compa ed o he gene al COSMO-RS scheme,
he COSMOquick app oach applied in his wo k o e ed accele a ed
compu a ion by assembling sigma p o iles o new molecules om a la ge
da abase o DFT esul s using a agmen -based me hod, as p e iously
desc ibed in he li e a u e (Loschen and Klam , 2012).
In line wi h he wo k low (Fig. 2), di e se equilib ium p ope ies and
comple e phase diag ams can be calcula ed. (Ecke and Klam , 2002;
Klam e al., 1998). The p esen s udy was making use o he calcula ed
(pseudo-) po en ials o ob ain a d ug o silica a ini y pa ame e in
p esence o wa e , which can be iewed as a pa i ioning coe icien
be ween a solu e and he ca ie su ace and wa e . In gene al, solu e
equilib a ion be ween any wo phases, P1 and P2, p o ides a pa i ion
coe icien K
(P1-P2)
acco ding o Eq. (2):
K(P1−P2)=exp(
μ
0P2−
μ
0P1
RT )(2)
whe e µ
0
is he chemical po en ial unde s anda d condi ions (wi h he
phase as a subsc ip ), R is he gas cons an , and T is he empe a u e. In
he p esen s udy, d ug pa i ioning be ween a silica su ace and an
aqueous bulk phase was o in e es . Fo p ac ical pu poses, he he -
modynamic calcula ions may assume he silica su ace as a supe cooled
bulk phase, and o he chemical po en ials,
μ
s
deno es a gi en chemical
(pseudo-) po en ial ha was ob ained by he desc ibed COSMOquick
app oach. Subsequen ly, a silica-wa e pa i ion coe icien K
si-w
was
es ima ed using Eq. (3):
ln(KSi−W) ≅ (
μ
S(wa e )−
μ
S(silica)
RT )(3)
2.2.7. S a is ical analysis and da a ep esen a ion
Co ela ion analyses we e pe o med using S a g aphics Cen u ion .
19.6.06 (S a g aphics Technologies Inc., Wa en on, USA). Co ela ion
analyses we e based on he calcula ed chemical po en ials and d ug
dissolu ion pe o mance pa ame e s, including he ini ial slope and AUC
alues. Linea Pea son co ela ion coe icien s we e calcula ed and
complemen ed wi h Spea man ank co ela ions ega ding he possible
occu ence o nonlinea i y o ou lie s. The ini ial dissolu ion slope was
ob ained om he linea ange (in e pola ion o he i s ew concen-
a ion poin s) o he elease cu es. Fo he ep esen a ion o g aphs and
calcula ion o AUC- alues (“quick ool o in eg a ion”) o each d ug
om he whole dissolu ion cu es (n =3), he so wa e O iginP o 2017
om O iginLab Co p. (No hamp on, USA) was used. O he calcula ions
we e based on Excel om he 365 apps o en e p ise package om
Mic oso Co p. (Redmond, USA). The s anda d de ia ions we e gene -
ally ob ained om expe imen s in iplica e unless s a ed o he wise.
3. Resul s and discussion
3.1. Manu ac u e and physical analysis o o mula ions
In he ea ly s ages o p e o mula ion, i is necessa y o assess he
po en ial o any o mula ion echnology o new d ug candida es.
The e o e, an in silico anking o d ug dissolu ion om mesopo ous silica
p oduc s would be highly aluable o pha maceu ical scien is s.
Cu en ly, he e a e some molecula simula ions, bu in a ull-a omis ic
e sion, hey can only ea compa a i ely small sys ems, especially
when a quan um-chemical app oach has been aken (Delle Piane e al.,
2014; Gignone e al., 2015). Mo eo e , he cu en ly a ailable simula-
ions in pha maceu ics ocus on bina y d ug-silica in e ac ions wi hou
conside ing an aqueous phase o d ug elease. COSMO-RS is in e es ing
in his con ex as i combines quan um-chemical calcula ions wi h s a-
is ical he modynamics o he su ace segmen s (Klam , 2005, 1995;
Klam e al., 1998). The p esen wo k ocused on he di e ence in he
calcula ed chemical po en ials no malized by RT, and in line wi h Eq.
(3), his is a d ug’s silica o wa e pa i ion coe icien . Al hough he
solid silica is he eby ea ed as a supe cooled liquid, which is a simpli-
ied app oxima ion, his app oach p o ides a p ac ical way o calcula e
chemical po en ials. Acco dingly, hese alues should be deemed as i s
app oxima ions ha can be sui able o anking o compounds ega ding
expe imen al d ug elease pe o mance. This conside a ion p o ided a
Fig. 2. Schema ic ep esen a ion o he COSMO-RS app oach using quan um chemis y (densi y unc ional heo y (DFT)) and s a is ical he modynamics (sigma
p o iles and po en ials) o ob ain molecula p ope ies. Fu he de ails abou his schema ic COSMO-RS app oach can be in e ed om he li e a u e (Klam , 2005;
Klam e al., 1998; Loschen and Klam , 2012).
A. Niede quell e al.
Eu opean Jou nal o Pha maceu ical Sciences 214 (2025) 107283
5
hypo hesis o he cu en s udy, and 10 di e en d ug-loaded meso-
po ous silica o mula ions we e p epa ed ia he incipien we ness
imp egna ion me hod. To s udy d ug deso p ion om silica su aces, he
d ug-loading concen a ion was kep in en ionally a he heo e ical
monolaye capaci y le el (MLC le el), and he mass gain du ing he
d ug-loading p ocess was g a ime ically e i ied o be close o he
a ge ed MLC alues (Table 1). Addi ionally, Table 2 shows he glass
o ming abili y classi ica ion (class I =c ys allize apidly om a mel , II
=c ys allize slowly om a mel and III ha do no c ys allize om a
mel upon e-hea ing and e-cooling (Bai d e al., 2010)), he
d ug-solubili y in he dissolu ion medium and he cha ge s a e a
physiological condi ions (+ = posi i ely cha ged, - =nega i ely cha ged
and 0 =neu al).
X- ay di ac ion (XRD) and di e en ial scanning calo ime y (DSC)
we e used o he basic d ug s a e cha ac e iza ion. As a esul , all
o mula ion samples we e p edominan ly amo phous. Possible aces o
c ys allini y can occu du ing he loading p ocess, bu his was no
deemed as c i ical o o mula ion pe o mance a ibu es as long as no
u he d ug c ys alliza ion occu s. In addi ion o d ug concen a ion,
he po e size dis ibu ion o he mesopo ous ca ie ma e ial, glass-
o ming abili y (GFA) o he d ug, and choice o sol en du ing d ug
loading can signi ican ly in luence he success o amo phous d ug
loading. The GFA desc ibes how eadily a compound can o m and
emain in he amo phous (glassy) s a e, which is a key ac o in
designing amo phous solid dispe sions (ASDs) (Augus ijns and B ews e ,
2012; Wy enbach and Kuen z, 2017). Good glass o me s (GFA class III)
emain amo phous on cooling and ehea ing, while in e media e glass
o me s (GFA class II) o m an amo phous glass bu ec ys allize o e
ime du ing s o age o upon ehea ing. Finally, poo glass o me s (GFA
class I) ypically c ys allize immedia ely upon cooling om a mel . I has
been desc ibed ha class II and I d ugs can ec ys allize du ing he
sol en -based loading p ocess, he eby blocking he mesopo es o he
ca ie so ha he d ug possibly emains a he su ace (Alhalaweh e al.,
2014; Di zinge e al., 2019; Wy enbach and Kuen z, 2017). The GFA o
he d ugs used is shown in Table 2. Fo diclo enac ( ee acid), i e -
mec in, and quinidine, he e was no in o ma ion abou he GFA a ail-
able in he li e a u e; he e o e, he ca ego iza ion was based on ou own
DSC measu emen s o he pu e d ug subs ance. Pimozide and quinidine
exhibi ed a c ys alliza ion and a mel ing peak du ing he second hea ing
and we e he e o e classi ied as GFA class II subs ances (see Table 2).
The choice o mos ly GFA III and class II compounds was o a oid d ug
c ys alliza ion issues du ing d ug loading as well as o enable a be e
compa ison o elease p o iles, ha a e po en ially less complica ed by
as d ug p ecipi a ion on elease (Di zinge e al., 2019).
3.2. Calcula ions o chemical po en ial-based silica-wa e pa i ioning
coe icien
To e alua e he pa i ioning beha io o selec ed APIs be ween a
silica su ace and an aqueous phase in silico (d ug a ini y o silica
du ing dissolu ion), he chemical po en ials in bo h en i onmen s we e
compu ed using COSMOquick, and he esul s a e lis ed in Table 3.
The chemical po en ial o he sel -associa ion o d ug molecules
(µs
(sel )
) con ibu es o solubili y bu no explici ly o d ug pa i ioning.
Mo eo e , a a heo e ical monolaye capaci y (MLC), d ug in e ac ions
wi h he silica su ace a e o highe impo ance han sel -in e ac ions o
d ug molecules, a leas in dilu ed aqueous condi ions. The las column
in Table 3 gi es he calcula ed silica-wa e pa i ioning coe icien in
line wi h Eq. (3), and no able di e ences be ween he compounds we e
ob ained. Highe alues in his column sugges a ela i ely g ea e
appa en d ug a ini y o silica in he p esence o wa e . Howe e , hese
calcula ions in ol e ce ain simpli ica ions. Mos no ably, silica is
assumed o be a supe cooled bulk phase. The e o e, he cu en s udy
aimed o de i e a pa ame e capable o a leas quali a i ely anking
di e en d ugs based on hei a ini y o silica du ing he d ug elease
p ocess.
Table 2
Glass o ming abili y (GFA), he equilib ium solubili y in he dissolu ion medium (phospha e bu e pH 6.8 wi h 0.5 % (w/ ) DOSS), and he physiological cha ge s a e.
Subs ance GFA class Solubili y [
μ
g/mL] Physiological cha ge s a e
F
Ca edilol III
A, B
36.1 ±0.01 +
Diclo enac ( ee acid) III
B
485.6 ±0.00 –
Indome hacin III
B, C, E
485.1 ±0.02 –
I e mec in III
B
4.8 ±0.01 0
Ke oconazole III
A, B, C, E
241.9 ±0.02 0
Lopina i III
B, D
34.4 ±0.03 0
Pimozide II
B
113.9 ±0.02 +
P obucol III
A, B, E
166.7 ±0.02 0
Quinidine II
B
1013.2 ±0.13 +
Ri ona i III
A, B, D, E
5.7 ±0.02 0
Sou ces: A: (Alhalaweh e al., 2014); B: measu ed 2 hea ing & 2 cooling cycles (indi idually wi hin a ange o 25 ◦C - 270 ◦C and 10 ◦C/min) wi h DSC; C: (Panini e al.,
2019); D: (Wy enbach and Kuen z, 2017); E: (Bai d e al., 2010); F: (Knox e al., 2024).
Table 3
Calcula ed chemical (pseudo-) po en ials o sel -associa ion (µs
(sel )
) and sol a ion (µs) be ween d ug-silica, d ug-wa e , and he silica-wa e pa i ioning coe icien o
d ug molecules (µs
(wa e )
- µs
(silica)
) / RT (acco ding o Eq. (3)) a 37 ◦C.
Subs ance name µs
(sel )
(kJ/mol) µs
(silica)
(kJ/mol) µs
(wa e )
(kJ/mol) (µs
(wa e )
- µs
(silica)
) / RT (-)
Ca edilol −14.44 −51.50 25.70 29.94
Diclo enac ( ee acid) −8.78 −33.31 19.84 20.61
Indome hacin −11.19 −39.53 19.82 23.02
I e mec in −62.78 −131.15 −10.42 46.82
Ke oconazole −22.26 −81.66 6.07 34.02
Lopina i −37.13 −83.29 13.63 37.59
Pimozide −27.39 −68.13 17.56 33.23
P obucol −43.81 −74.65 27.38 39.57
Quinidine −12.07 −43.24 17.60 23.59
Ri ona i −34.81 −108.53 5.58 44.25
A. Niede quell e al.
Eu opean Jou nal o Pha maceu ical Sciences 214 (2025) 107283
6
3.3. Expe imen al in i o dissolu ion es ing o d ug-loaded silica
o mula ions
T adi ional dissolu ion es ing is impo an o o mula ion de el-
opmen and cha ac e iza ion o bio-enabling d ug deli e y sys ems, like
o example, supe sa u a ing mesopo ous silica-based o mula ions. I is
impo an o conside ha he design and execu ion o he dissolu ion
es can s ongly in luence he ou comes o he s udy (Augus ijns and
B ews e , 2012). In his s udy, dissolu ion es ing was pe o med using a
USP Appa a us II (paddle me hod) wi h 250 mL o phospha e bu e (pH
6.8) con aining 0.5 % (w/ ) dioc yl sodium sul osuccina e (DOSS) as he
dissolu ion medium. All expe imen s we e conduc ed a 37 ±0.5 ◦C.
Fig. 3 shows he cou se o he cumula i e concen a ions o he 10
pha maceu ical d ugs. In all cases, d ug dissolu ion om he meso-
po ous silica o mula ions was incomple e a e 3 h o obse a ion,
which aligns wi h s ong d ug-silica in e ac ions and po en ial
e-adso p ion p ocesses, a he han la e p ecipi a ion, as indica ed by
he s abili y o pla eau concen a ions h oughou he obse a ion
pe iod. The expe imen al design — inco po a ing monolaye d ug
loading o minimize d ug-d ug e ec s and a medium olume (250 mL)
ha ansi ioned om sink o non-sink condi ions depending on solu-
bili y — was hus well-sui ed o in es iga ing d ug-silica in e ac ions
unde con olled condi ions. Among all es ed APIs, quinidine exhibi ed
he highes d ug elease a e 3 h, wi h a elease ac ion o app oxi-
ma ely 62 %. This, a physiological condi ions posi i ely cha ged d ug
subs ance, also showed he highes solubili y in he dissolu ion media
(see Table 2). In con as , ke oconazole, i ona i , and i e mec in
demons a ed he lowes d ug elease o e he same pe iod, wi h a
eleased ac ion o app oxima ely 22 %, 27 %, and 29 %, espec i ely.
Ri ona i and i e mec in also possessed he lowes solubili y in he
dissolu ion media (Table 2). No ably, wi hin he i s wo minu es o he
dissolu ion p ocess, a leas 7 % o he d ug was eleased ac oss all
o mula ions. The mos p onounced inc ease in d ug elease occu ed
wi hin he ini ial 5–10 min o all es ed subs ances. Subsequen ly, he
cumula i e d ug elease a e slowed down conside ably. A e 90 min, a
pla eau o mos d ugs was eached, a e which he d ug concen a ion
in he dissolu ion medium emained cons an . Only ca edilol, diclo e-
nac ( ee acid), indome hacin, and quinidine showed, in compa ison o
he beginning, a slowe bu con inuous inc ease in d ug dissolu ion ill
he end.
Theo e ically, he s eepness o he ini ial elease slope would be o
some ex en de e mined by he d ug-silica in e ac ions. Howe e ,
u he molecula p ope ies can be in luen ial o d ug elease, so a high
sp ing e ec can, o example, a ec he ini ial inc ease o d ug
concen a ions (Sun and Lee, 2013). As o such a sp ing e ec and
hence he p opensi y o a d ug o supe sa u a e, i was shown p e iously
o g ea ly depend on GFA (Blaabje g e al., 2018). As he model d ugs in
he cu en s udy we e mos ly good glass- o ming compounds (Table 2),
he e ec o supe sa u a ion p opensi y was no expec ed o domina e
he o e all d ug elease. Howe e , i is s ill expec ed ha such
d ug-speci ic p ope ies beyond appa en d ug a ini y o silica would
a ec he kine ics o d ug elease om mesopo ous o mula ions.
Based on Fig. 3, he pe o mance pa ame e s (ini ial slopes and AUC
alues) we e ob ained. The supe sa u a ion a io was de e mined om
he solubili y-no malized concen a ions a he end o he dissolu ion
p ocess, a e aged om he las h ee measu ing poin s o he pla eau.
Table 4 lis s he esul s ob ained o he h ee pa ame e s o he selec ed
10 ac i e pha maceu ical d ugs (APIs). These calcula ed a ios indica e
he inal sa u a ion le el, whe e alues <1 ep esen unde sa u a ed
solu ions, alues equal o o close o 1 ep esen sa u a ed solu ions, and
alues >1 ep esen supe sa u a ed solu ions.
O e all, i e mec in, i ona i , ca edilol, and lopina i showed he
highes supe sa u a ion le els a he end o he dissolu ion expe imen s,
as e lec ed by he pla eau egion a he end o hei espec i e disso-
lu ion cu es. The as es ini ial dissolu ion, as indica ed by he slope
alues wi hin he i s 5–10 min, was obse ed o quinidine, ca edilol,
and ke oconazole.
Rema kably, i e mec in and i ona i we e supe sa u a ed du ing
he en i e dissolu ion expe imen , while ha ing he lowes solubili y
alues in he dissolu ion media compa ed o he o he d ug subs ances
(Tables 2 and 4). Ca edilol and lopina i showed le els o a sa u a ed
solu ion s a e, whe eas he es o he d ugs (diclo enac ( ee acid),
indome hacin, ke oconazole, pimozide, p obucol, and quinidine) we e
unde sa u a ed du ing elease kine ics (see Table 4). Mo eo e , quini-
dine, ca edilol, and ke oconazole exhibi ed he as es ini ial dissolu-
ion a es, whe eas quinidine, pimozide, and indome hacin
demons a ed he highes a ea unde he cu e (AUC) alues a e 3 h.
The esul s indica e ha di e ences in ela i e sa u a ion le els may
con ibu e o a po en ial mechanism ha can be con ounded wi h
appa en d ug-ca ie in e ac ions, ul ima ely in luencing d ug elease
kine ics. Howe e , om a p ac ical pe spec i e, i is common o mo e
han one mechanism o go e n he pe o mance o a supe sa u a ing
o mula ion, ei he in i o o in i o (O’Dwye e al., 2019).
3.4. Co ela ion analysis – chemical po en ials as a p edic i e ool o
quali a i e dissolu ion pe o mance?
As men ioned be o e, he calcula ed silica-wa e pa i ioning
Fig. 3. Dissolu ion cu es o 10 d ug-loaded mesopo ous silica o mula ions (n
=3) o e 3 h a 37 ◦C in phospha e bu e (pH 6.8) and 0.5 % DOSS.
Table 4
Pe o mance pa ame e s o d ug elease kine ics (i.e., ini ial slopes and AUC
alues o d ug concen a ions 0–180 min) and supe sa u a ion a io a he end o
he dissolu ion p ocess (means o he las h ee poin s) o 10 d ug-loaded mes-
opo ous silica o mula ions (n =3) a 37 ◦C.
Subs ance
name
Ini ial slope o
d ug elease
[µg⋅mL⁻¹⋅min⁻¹]
AUC o
concen a ions
[µg⋅min⋅mL⁻¹]
Supe sa u a ion
a io
1
a he end o
he expe imen (180
min) [-]
Ca edilol 2.16 ±0.09 6209±349 1.11 ±0.04
Diclo enac
( ee acid)
0.35 ±0.05 5735±434 0.074 ±0.00
Indome hacin 1.02 ±0.01 7745 ±291 0.10 ±0.00
I e mec in 1.16 ±0.12 4911 ±374 6.33 ±0.11
Ke oconazole 1.85 ±0.15 5220 ±829 0.13 ±0.00
Lopina i 1.67 ±0.12 6154 ±514 1.09 ±0.03
Pimozide 1.77 ±0.17 8840 ±370 0.46 ±0.00
P obucol 1.53 ±0.11 6195 ±134 0.23 ±0.00
Quinidine 2.86 ±0.15 9622 ±746 0.06 ±0.00
Ri ona i 1.42 ±0.18 4506 ±525 4.77 ±0.11
1: Supe sa u a ion a io: (C/C
sa
, whe e C is he d ug concen a ion and C
sa
is
he equilib ium solubili y).
A. Niede quell e al.
Eu opean Jou nal o Pha maceu ical Sciences 214 (2025) 107283
7
coe icien was coined as a po en ial cha ac e is ic o appa en d ug-
silica a ini y o co ela e wi h expe imen ally de e mined dissolu ion
pe o mance. A simpli ying assump ion he e was ea ing silica as a
supe cooled bulk liquid. A simila assump ion has been p e iously
made, pa icula ly when conside ing he solubili y pa ame e s o d ugs
in a ious sol a ion and solubili y- ela ed s udies (Alanazi e al., 2020;
Alqa ni e al., 2021; Kalam e al., 2019; Shakeel e al., 2020; Shakeel and
Alsheh i, 2020). COSMO-RS is he e ce ainly a mo e ad anced way o
model he modynamic p ope ies o d ugs and hei excipien mix u es,
and i add esses se e al o he sho comings o solubili y pa ame e s ha
we e discussed be o e in he li e a u e (Janko ic e al., 2019).
The silica-wa e pa i ion coe icien , acco ding o Eq. (3), was
compa ed wi h he d ug elease cha ac e is ics. Thus, Figs. 4 and 5
display he linea sha ed co ela ions in sca e plo s be ween he ini ial
loga i hmic slope o he a ea unde he cu e alues (AUC un il 180
min) on he y-axis and he silica-wa e pa i ioning coe icien ((µs
(wa e )
- µs
(silica)
) / RT) o he d ugs on he x-axis. Fo he co ela ion analysis,
only eigh d ugs we e conside ed, while diclo enac ( ee acid) and
indome hacin we e ea ed as ou lie s. These wo pha maceu ical ac i e
ing edien s (APIs) belong o he g oup o nons e oidal an i-
in lamma o y, small acidic d ugs. Bo h ha e ca boxyl g oups and exis
wi h pKa alues o 4.0 and 4.5, mos ly in a dep o ona ed/ionized o m a
physiological pH. Bo h showed a e y low appa en a ini y o silica
(low silica-wa e pa i ioning coe icien ), and excep o quinidine,
which had an e en highe solubili y, hese wo d ugs showed he highes
solubili y in he phospha e bu e a pH 6.8. The ela i ely high solu-
bili y combined wi h he low ca ie a ini y was likely a key ac o in he
la e a ini y no being p ominen ly obse ed in he co ela ion analysis
o he elease pe o mance o he wo d ugs. Besides exhibi ing unusual
s uden ized esiduals (wi h alues <3) in he eg ession model, diclo-
enac is known o sel -associa e in aqueous media; he acid unc ion
in e ac s wi h i sel and o ms dime s (Kozlowska e al., 2018). I con-
ains wo a oma ic ings and a chlo ine-subs i u ed, hyd ophobic co e.
The ca boxyla e g oup makes i anionic, bu he es o he molecule is
s ongly hyd ophobic, which p omo es agg ega ion. Owing o his ag-
g ega ion, he d ug deso p ion kine ics (ini ial slope o d ug concen-
a ion) we e appa en ly educed. Likewise, indome hacin is known o
unde go sel -associa ion h ough he o ma ion o hyd ogen-bonded
ca boxylic acid dime s in solu ion (Taylo and Zog a i, 1997), which
can a enua e i s appa en dissolu ion beha io . The e o e, he obse ed
unusual s uden ized esiduals o diclo enac and indome hacin in he
eg ession analysis we e a ibu ed p ima ily o such sel -agg ega ion.
Mo eo e , ionizable d ug acids ha exhibi ela i ely high solubili y
a an in es inal pH would no make bes candida es o bio-enabling
o mula ions, such as mesopo ous silica-based d ug deli e y sys ems
(Amidon e al., 1995; Cha alabidis e al., 2019). The same a gumen s
hold ue o quinidine, which is classi ied as biopha maceu ical clas-
si ica ion sys em (BCS) class I d ug (Mo i e al., 2012), whe eas he o he
model compounds we e d ugs o class II (Busca ini e al., 2024; Chua-
suwan e al., 2009; Fu e al., 2019; Guo e al., 2014; Hamed e al., 2016;
Sodei ian e al., 2021; S eenekamp e al., 2024) and IV (Saeed e al.,
2021; Salim e al., 2023). O e all, ega ding he co ela ion analysis, he
loga i hmic ini ial slope alues exhibi ed a s ong linea Pea son co -
ela ion wi h he silica-wa e pa i ioning coe icien ( =–0.98, p <
0.0001) and a Spea man ank co ela ion o =–0.98 (p <0.01). The
loga i hmic AUC alues also co ela ed wi h he silica-wa e pa i ion-
ing coe icien , wi h a Pea son co ela ion o =–0.79 (p =0.02) and a
Spea man ank co ela ion o =–0.86 (p =0.02). Addi ionally, when
he loga i hmic ini ial slope o he pe cen age o nominal d ug eleased
was conside ed, a Pea son co ela ion o =–0.86 (p =0.006) and a
Spea man ank co ela ion o =0.81 (p =0.03) wi h he silica-wa e
pa i ioning coe icien we e obse ed. These s ong co ela ion e-
la ionships, illus a ed in Figs. 4 and 5, indica e ha a highe
silica-wa e pa i ioning coe icien e lec s a g ea e a ini y o he d ug
o he silica su ace and, consequen ly, a lowe ini ial slope o d ug
elease in o he dissolu ion medium.
Simila ly, Fig. 5 illus a es ha a g ea e a ini y o silica co esponds
o a lowe a ea unde he cu e (AUC) o he dissol ed d ug concen-
a ions o e he en i e dissolu ion pe iod (180 min). Howe e , as
indome hacin and diclo enac ( ee acid) we e he only nega i ely
cha ged d ugs (Table 2), hei elease kine ics we e no expec ed o
depend signi ican ly on silica in e ac ion unde ionizing pH condi ions,
as men ioned p e iously. Ionizable acids a e expec ed o weakly in e ac
wi h silica because o epulsi e o ces. Howe e , as discussed ea lie ,
d ug agg ega ion ia he ca boxyla e g oups could ha e g ea ly educed
such epulsi e o ces in he expe imen s. As he in silico model was no
aking such agg ega es in o accoun , his would explain he de ia ion
om he gene al endline. The opic o elec os a ic in e ac ions is o
b oade in e es . In his ega d, p e ious wo k (Ha e e al., 2020a)
in es iga ed he e ec s o elec os a ic in e ac ions be ween d ug mol-
ecules and silica su ace on he dissolu ion o weakly basic d ugs om
mesopo ous silica-based o mula ions as a unc ion o medium pH. The
au ho s concluded om hei indings ha , compa ed o he o ma ion
o hyd ogen bonding in e ac ions, d ug-silica elec os a ic in e ac ions
lead o highe adso p ion and lowe d ug elease. The expe imen al da a
Fig. 4. Ini ial slope o d ug elease e sus silica-wa e pa i ion coe icien plo
o eigh d ugs (and wo ou lie s) wi h co ela ion coe icien s. By de ini ion,
he slope was ini ially exp essed in he ollowing uni : [µg⋅mL
-1
⋅min
-1
].
Fig. 5. A ea unde he cu e (AUC) alues e sus silica-wa e pa i ion coe -
icien plo o eigh d ugs wi h co ela ion coe icien s. The AUC alues we e
ini ially exp essed in he ollowing uni : [(µg⋅min)⋅mL
-1
].
A. Niede quell e al.
Eu opean Jou nal o Pha maceu ical Sciences 214 (2025) 107283
8
in he p esen wo k suppo s such a iew, ha s ong in e ac ions wi h
silica limi d ug elease, bu also sugges s ha elec os a ic in e ac ions
alone do no ully accoun o he obse ed deso p ion beha io . Fo
example, ca edilol, a weak base wi h a pKa o app oxima ely 7.8 is
p ima ily p o ona ed and hus posi i ely cha ged a pH 6.8 (Yu a aja
and Khanam, 2014). Al hough his leads o a ac i e elec os a ic in-
e ac ions be ween he d ug and he nega i ely cha ged silica su ace,
ca edilol showed a ela i ely low silica wa e pa i ioning coe icien
and he e o e limi ed d ug-silica a ini y (Figs. 4 and 5). The complexi y
o he dissolu ion p ocess o a d ug-loaded mesopo ous silica o mula-
ion can be la gely a ibu ed o he in e play o mo e han one
con ibu ing ac o ; he e o e, he elease kine ics o any compound a e
no go e ned solely by he cha ge s a e o he d ug a he pH o he
dissolu ion medium.
In Fig. 5, he p e iously iden i ied “ou lie s” o he ini ial slope o
d ug concen a ions, diclo enac ( ee acid) and indome hacin, a e less
dis an om he da a cloud o he o he d ugs. They we e no included in
he co ela ion analysis o consis ency and hence, compa abili y o he
analysis in Fig. 4. While his p e ious analysis ocused on he ini ial
s age o he d ug elease p ocess, he AUC alues a e ep esen a i e o
he d ug dissolu ion p ocess o e he en i e expe imen al ime (3 h).
Du ing his long dissolu ion ime, o he in luen ial ac o s besides d ug
deso p ion om silica can be in luen ial ega ding he dissolu ion pe -
o mance. In gene al, i is expec ed ha he physicochemical p ope ies
ha go e n solubili y and supe sa u a ion p opensi y in luence kine ic
p o iles and ha e a highe impac on he AUC o he en i e expe imen
du a ion compa ed o he sho ime o he ini ial elease. Howe e ,
acco ding o he dissolu ion p o iles in Fig. 3, he d ug concen a ion
pla eaus we e no all le eling o du ing he 3 h obse a ion ime. The e
we e no dec eases in concen a ions obse ed ha could ha e o he wise
indica ed a ma ked d ug p ecipi a ion o d ugs ha we e
supe sa u a ed.
No ably, quinidine exhibi ed ex eme beha io ac oss se e al d ug
p ope ies, including he highes solubili y in he dissolu ion medium
(lowes supe sa u a ion a io), highes ini ial d ug elease and AUC
pla eau alues, and low a ini y o silica. In con as , i e mec in had he
lowes solubili y in he dissolu ion medium (highes supe sa u a ion
a io), a medium ini ial slope o d ug concen a ion, low AUC le el, and
highes a ini y o silica. In summa y, he elease p ocess om meso-
po ous silica appea s o be go e ned by a complex in e play o mul iple
ac o s; howe e , he appa en d ug-silica a ini y plays a dominan ole,
especially in he ini ial s age o d ug elease. O e all, i is e y
encou aging ha , o neu al and posi i ely cha ged d ugs, he e was a
e y good co ela ion ound be ween he silica-wa e pa i ioning co-
e icien (based on chemical po en ials) and he expe imen al pe o -
mance pa ame e s o d ug elease (ini ial slopes and AUC alues). S ill,
despi e ex ended obse a ion o e 3 h, none o he es ed ac i e pha -
maceu ical ing edien s (APIs) achie ed comple e elease, sugges ing
ha a ac ion o d ug molecules emained adso bed o he mesopo ous
silica su ace. This pe sis en e en ion likely e lec s a dynamic equi-
lib ium be ween deso p ion and e-adso p ion p ocesses, as p e iously
obse ed in he li e a u e (Tu ku e al., 2007). Ano he s udy (McCa hy
e al., 2018) u he showed ha he ex en o i e e sible adso p ion is
highly pH-dependen . Gi en he isoelec ic poin o silica (~pH 2), i s
su ace bea s a signi ican nega i e cha ge a physiological and highe
pH le els, wi h cha ge densi y inc easing ma kedly be ween pH 6 and 11
(A kin e al., 2003; Sha ma e al., 1996), and his p omo es in pa icula
elec os a ic in e ac ions.
Incomple e elease om mesopo ous sys ems is a widely epo ed
phenomenon, e en a low d ug loadings (Le e al., 2019), and is
commonly a ibu ed o e e sible and i e e sible d ug-silica in-
e ac ions (Ha e e al., 2020a). Consis en wi h his, he li e a u e de-
sc ibes cases o incomple e elease o mos e alua ed d ugs in his s udy
(i.e., i ona i and lopina i (Dening e al., 2019), indome hacin
(Ga cia-Benne e al., 2018; McCa hy e al., 2020), ke oconazole (Ha e
e al., 2020a), i e mec in (Velho e al., 2024), ca edilol (Mahdi e al.,
2022), diclo enac (Ghio ghi a e al., 2022), p obucol (Lau e al., 2019)).
I sink condi ions o a di usion ba ie (memb ane) a e p esen , elease
may app oach comple ion be e , as he e would be a sink o he d ug o
di use away. Ha e e al. (Ha e e al., 2020a) e.g., showed ha a azana i
e en ually eached 100 % elease when an abso p i e sink was used.
The u ili y o COSMO-RS chemical po en ials o app oxima e he
silica-wa e pa i ioning beha io , and hei s ong co ela ion wi h
d ug elease me ics (i.e., ini ial slope and AUC), p o ides he means o
a leas ank o mula ion elease pe o mance. This is encou aging o
o mula ion scien is s, who can ha ness his app oach o o mula ion
de elopmen .
In he u u e, mo e expe imen s wi h di e en d ug candida es,
highe d ug loadings, and di e en aqueous su ounding dissolu ion
media should be pe o med o a mo e comple e unde s anding o he
molecula d ug-silica in e ac ions du ing he complex elease p ocess.
4. Conclusions
This s udy in oduces a no el in silico me hod o in es iga ing d ug-
silica in e ac ions in aqueous en i onmen s, add essing a c i ical
knowledge gap in he a ional design o mesopo ous silica-based de-
li e y sys ems o poo ly soluble d ugs. Using a silica-wa e d ug pa -
i ioning coe icien om COSMO-RS calcula ions, good linea and ank
co ela ions wi h he expe imen al d ug elease cha ac e is ics we e
ob ained. The indings o his s udy unde sco e he impo ance o
conside ing d ug-speci ic physicochemical p ope ies, such as ioniza ion
s a e, solubili y, and glass- o ming abili y, in andem wi h he su ace
cha ac e is ics o he mesopo ous silica ca ie (su ace a ea, po e size,
and silanol-siloxane a io). No ably, while elec os a ic in e ac ions
con ibu e subs an ially o he d ug a ini y o he silica su ace, hey do
no solely go e n he deso p ion p ocess. The esul s suppo he iew
ha d ug elease kine ics om mesopo ous silica sys ems a e no
dic a ed by a single dominan in e ac ion bu a he by he collec i e
e ec o mul iple, o en compe ing, molecula mechanisms.
The cu en s udy compa ed d ugs ela i e o hei abili y o o m a
monolaye on he ca ie su ace. A such low d ug loadings nea he
heo e ical monolaye capaci y (MLC), he dominan in e ac ions occu
a he solid-liquid in e ace and in ol e a complex balance o elec o-
s a ic a ac ion o epulsion (Ha e e al., 2020a, 2020b), hyd ogen
bonding, and dispe sion ( an de Waals) o ces, sugges ing a pe sis en
equilib ium be ween deso p ion and e-adso p ion, and highligh ing he
po en ial o i e e sible binding phenomena (McCa hy e al., 2018,
2016). The p esen ed in silico app oach o e ed a aluable heo y-based
ool o a leas quali a i ely ank d ug elease pe o mance. This would
be highly aluable in a pha maceu ical p o iling phase whe e minimal
amoun s o d ug a e a ailable o guide o mula ion de elopmen . By
enabling an ea ly mechanism-based anking o d ug-silica in e ac ions,
his me hod also has he po en ial o educe he expe imen al sc eening
ime in p e o mula ion p ac ice.
While his s udy ocused on GFA class III d ugs wi h low d ug loading
o enable a clea mechanis ic in e p e a ion o silica-wa e pa i ioning
e ec s, u u e in es iga ions may ex end his app oach o include GFA
class I compounds. Howe e , due o hei poo amo phous s abili y and
high c ys alliza ion endency, such compounds would equi e speci -
ically adap ed o mula ion s a egies, such as su ace unc ionaliza ion
o he ca ie o co- o mula ion wi h c ys alliza ion inhibi o s. In a
b oade con ex , u u e wo k should also explo e di e en d ug loadings
and dissolu ion media a a ying pH alues o gain u he insigh in o
how hese ac o s in luence d ug-silica in e ac ions and ul ima ely d ug
elease. Mo eo e , he de elopmen o quan i a i e models o d ug
elease may bene i om inco po a ing addi ional physicochemical pa-
ame e s o he ca ie ma e ial. Howe e , he p esen s udy al eady
con ibu es o ad ancing he mechanis ic unde s anding o appa en
d ug-silica in e ac ions, he eby suppo ing he a ional design o mes-
opo ous silica-based d ug deli e y sys ems.
A. Niede quell e al.
Eu opean Jou nal o Pha maceu ical Sciences 214 (2025) 107283
9
