`
`
`
`
`
`
`
`ELSEVIER
`
`International Journal of Pharmaceutics 212.(2001) 29-40
`
`
`
`international
`journal of
`pharmaceutics
`
`www.elsevier.com/locate/ijpharm
`
`Cyclodextrin solubilization of benzodiazepines: formulation
`of midazolam nasal spray
`
`T. Loftsson **, H. Guémundsdottir >, J.F. Sigurjonsdéttir *, H.H. Sigurdsson?,
`S.D. Sigfisson *, M. Masson *, E. Stefansson °
`" Faculty of Pharmacy, University of Iceland, P.O. Box 7210, IS-127 Reykjavik, Iceland
`> Department of Ophthalmology, University of Iceland, Landspitali (the University Hospital), IS-101 Reykjavik, Iceland
`
`Received 9 May 2000; received in revised form 15 September 2000; accepted [9 September 2000
`
`Abstract
`
`The cyclodextrin solubilization of three benzodiazepines, ic. alprazolam, midazolam and triazolam, was investi-
`gated. The cyclodextrin solubilization was enhanced through ring-opening of the benzodiazepine rings and ionization
`of the ring-open forms. Additional enhancement was obtained through interaction of a water-soluble polymer with
`‘the cyclodextrin complexes. The ring-opening was pH-dependent and completely reversible, the ring-open forms
`dominating at low pH but the ring-closed forms at physiologic pH. The ring-closed forms were rapidly regenerated
`upon elevation of pH. In freshly collected human serum in vitro at 37°C, the half-life for the first-order rate constant
`for the ring-closing reaction was estimatedtto be less than 2min for both alprazolam and midazolam. Midazolam (17
`
`mg/ml) was so
`ze
`14%.(v/v) sulfobutylether B-cyclodextrin,
`0.1% (w/v) hye
`ypropy!‘methylcellulose, preservatives andbuffersalts.‘Six healthy volunteers received0.06:1g/kg
`
`midazolam intranasally and 2 mg intravenously, and blood samples were collected up to 360 min after the
`‘administration. Midazolam was absorbed rapidly reaching maximum serum concentrations of54:3% 5.0 ng/mi at
`
`e2 min. The elimination half-life of midazolam was 2.2+0.3 bh and the absolute availability was?737%. All
`mean values +SEM. © 2001 Elsevier Science B.V. Al! rights reserved.
`
`Keywords: Solubility; Benzodiazepines; Cyclodextrin, Complexation; Ionization
`
`1. Introduction
`
`Most benzodiazepine drugs are derivatives of
`2,3-dihydro-1H-1,4-benzodiazepine with sedative,
`antianxiety, anticonvulsant and muscle relaxant
`
`* Corresponding author. Tel.: + 354-5254464; fax: + 354-
`5254071.
`E-mail address: thorstlo@hi.is (T. Loftsson).
`
`properties. In pharmaceutical formulations benzo-
`diazepines are mainly used as the solid base, and
`as such they are readily dissolved in lipophilic
`solvents or in polar organic solvents such as etha-
`nol. Formulation of the benzodiazepine bases in
`aqueous drug formulation has been hampered by
`their low aqueous solubility and frequently the
`:
`ifpharmaceuti-
`
`only .practical..zmeans--of» obtaitiin;
`“solutions
`is
`cally.....acceptable.... benzodiazepine™
`
`0378-5173/01/% - see front matter © 2001 Elsevier Science B.V. All rights reserved.
`PIT: 80378-5173(00)00580-9
`
`AQUESTIVE EXHIBIT 1125 Page 0001
`AQUESTIVE EXHIBIT 1125 Page 0001
`
`
`

`

`
`
`30
`
`T. Loftsson et ai. /International Journal of Pharmaceutics 212 (2001) 29-40
`
`through-:the-use~ofcombinations. of cosolvents
`Bc
`1997; Alvarez Nuiifies and
`Yalkowsky, 1998)."Unfortunately, administration
`of-non-aqueous-drug.formulations may. result in
`pain,.-irritation..and.drug. precipitation upon ad-
`ministration (Yalkowsky and Rubino, 1985; Way
`and Brazeau, 1999). Replacing the cosolvent for-
`mulations with aqueous cyclodextrin containing
`drug formulations may circumvent these side ef-
`fects (Brewster et al., 1989; Brewster, 1991; Brew-
`ster
`and
`Loftsson,
`1999).
`Previously,
`benzodiazepines have been solubilized through cy-
`clodextrin complexation (Kraus et al.,
`1991;
`Lofisson et al., 1994). However, the complexation
`efficacy is frequently low and,
`thus,
`relatively
`large amounts of cyclodextrin are needed to solu-
`bilize small amounts of a given benzodiazepine
`drug. Increased complexation efficacy can be ob-
`tained by increasing either the intrinsic solubility
`of the drug (S,) or the apparent stability constant
`(K,) of the drug/cyclodextrin complex, or by in-
`creasing both simultaneously (Loftsson,
`1998;
`Loftsson etal., 1999). In aqueous solutions, some
`drugs can exist in more than one structural form,
`e.g. equilibrium isomers or ionization stages. Al-
`though the individual forms are in equilibrium
`with each other, and thus not totally independent
`_of each other, the overall aqueous solubility (or
`apparent S,) of a drug can be enhanced through
`formation of such multiple structural forms.
`Cyclic imines, such as 2,3-dihydro-1H-1,4-ben-
`zodiazepine, are known to undergo reversible and
`pH-dependent ring-opening through formation of
`aldehyde or ketone and a primary amine:
`
`
`
`COC
`
`CHO
`
`NH;*
`
`Under certain conditions open and closed
`forms are both present in aqueous solutions. Co-
`existence of such forms increases the apparent
`solubility of the benzodiazepine. Often, the ring-
`open form is an intermediate which is formed
`during benzodiazepine degradation in aqueousso-
`lutions but in some cases, e.g.
`in the case of
`alprazolam, midazolam and triazolam, this form
`
`.
`
`is chemically stable and can contribute to the
`overall aqueous solubility of the drug (Choet al.,
`1983; Kanto,
`1985; Kurono et al,
`1985;
`Kuwayama et al., 1986; Sbarbati Nudelman and
`de Waisbaum, 1995). For example, in the com-
`mercial aqueous intravenous(i.v.) solution of mi-
`dazolam (Dormicum®, Hoffmann-La Roche,
`Switzerland) the drug is 15-20% in the ring-open
`form and the pH is approximately 3.3 (Gerecke,
`1983). In addition, both forms, Le. the ring-open
`and the ring-closed midazolam, can exist in sev-
`eral different ionization forms. In the aqueousi.v.
`solution, the ring-open form of midazolam can be
`characterized as a midazolam prodrug since the
`ring is completely closed when the pH is elevated
`to 7.4. Previously we have shown that low com-
`plexation efficiency can hamper the usage of cy-
`clodextrins
`in
`certain
`pharmaceutical
`formulations and that both drug ionization and
`water-soluble polymers can enhance the complex-
`ation efficiency (Loftsson, 1998; Loftsson et al.,
`1999). Ionization of a drug molecule increases the
`apparent S, and addition of a water-soluble poly-
`mer to the complexation media increases K,.
`Several investigators have attempted to use the
`commercially available aqueous iv. solution for
`intranasal
`(i-n.) administration of midazolam
`(Bjérkman et al., 1997; Burstein et al., 1997). The
`midazolam concentration in this solution is only 5
`mg/ml. Thus,
`relatively large amounts of the
`acidic i.v, solution have to be sprayed into the
`nose in order to induce sedation and anxiolysis.
`Subsequently midazolam is only partly absorbed
`from the nasal cavity and partly from the gas-
`trointestinal tract after swallowing. The midazo-
`lam bioavailability after in. administration of the
`iv. solution is frequently about 50% (Burstein et
`al., 1997). To reduce spilling and swallowing of
`the iv. solution after in. administration, and to
`improve the bioavailability, the dosage has to be
`sprayed in small aliquots into the nasal cavity
`(Bjérkman et al., 1997). However, in. administra-
`tion of the acidic i.v. solution can cause severe
`irritation in the nasal cavity.
`The purpose of the present study was to investi-
`gate the effects of the reversible ring-opening of
`the diazepine ring and ionization on the cyclodex-
`trin complexation of benzodiazepines, as well as
`
`
`
`
`
`AQUESTIVE EXHIBIT 1125 Page 0002
`AQUESTIVE EXHIBIT 1125 Page 0002
`
`
`

`

`
`
`
`T. Laftsson et al. / International Journal of Pharmaceutics 212 (2001) 29-40
`
`31
`
`formulation and testing of physiologically accept-
`able aqueous midazolam nasal spray solution,
`
`2. Materials and methods
`
`2.1. Materials
`
`Midazolam base was purchased from Sifa
`(Shannon, Ireland), and alprazolam and triazolam
`from Sigma (St Louis, MO). Sulfobutylether-B-cy-
`clodextrin sodium salt with molar substitution of
`6.2 (Captisol®, SBEBCD) was kindiy donated by
`CyDex (Kansas City, KS). Randomly methylated
`B-cyclodextrin with degree of substitution (DS) of
`1.8 (RMBCD) and 2-hydroxypropyl-B-cyclodex-
`trin with DS of 0.6 (HPBCD) were kindly donated
`by Wacker-Chemie (Burghausen, Germany). Hy-
`droxypropy! methylcellulose 4000 (HPMC) was
`purchased from Mecobenzon (Denmark). All
`other chemicals used were of pharmaceutical or
`special analytical grade.
`
`2.2. Solubility studies
`
`An excess amount of the drug to be tested was
`added to water or aqueous Teorell—Stenhagen
`buffer system (Bates and Paabo, 1989), or the
`aqueous nasal
`formulation, containing various
`amounts of the different cyclodextrins with or
`without a polymer. The suspension formed was
`heated in an autoclave in a sealed container to
`130°C for at least 30 min. After cooling to room
`temperature (22--23°C) a small amount of solid
`drug was added to the container to promote
`precipitation. Then the suspension was allowed to
`equilibrate for at least 3 days at room tempera-
`ture, protected from light. After equilibration was
`attained, an aliquot of the suspension wasfiltered
`through a 0.45-y~m membranefilter (cellulose ace-
`tate from Schleicher & Schuell, Germany}, diluted
`with the HPLC mobile phase and analyzed by
`HPLC. The pH values reported were determined
`at room temperature at the end of the equilibra-
`tion period.
`The effect of pH on the stability constant (K,)
`of the drug/cyclodextrin (1:1) complex was deter-
`mined as previously described (Loftsson and Pe-
`
`tersen, 1998), Briefly the drug solubility was
`determined in aqueous nasal formulation contain-
`ing from 0 to 14% (w/v) cyclodextrin. The compo-
`sition of the nasal formulation was as follows:
`benzalkonium chloride
`(0.02% w/v), EDTA
`(sodium edetate) (0.1% w/v}, HPMC (0.1% w/v},
`phosphoric acid (0.43% v/v) and aqueous sodium
`hydroxide solution (for pH adjustment) in water.
`As before,
`the exact pH of each solution was
`determined at the end of the equilibration period.
`Differences in pH were corrected by drawing the
`pH-solubility profiles at each cyclodextrin concen-
`tration and determining the solubilities of the
`drug from these profiles at selected pH values.
`The values obtained were used to draw the phase-
`solubility diagrams, all of which were linear, Fi-
`nally, K, was calculated from the equation
`(Higuchi and Connors, 1965):
`_
`Slope
`° S,(1 — Slope)
`where K, is the stability constant of the drug—cy-
`clodextrin (1:1) complex, slope is the calculated
`slope of the linear phase-solubility diagram and
`5, is the apparent intrinsic solubility of the free
`drug determined in the aqueous complexation me-
`dia, at appropriate pH, when no cyclodextrin or
`polymer was present.
`
`2.3, Quantitative determinations
`
`The quantitative determination of drugs was
`carried out on a high performance liquid chro-
`matographic (HPLC) component systern consist-
`ing of ConstaMetric
`3200
`isocratic
`solvent
`delivery system operated at 1.50 ml/min, a Merck-
`Hitachi AS4000 autosampler, a Luna Cj, 5 pm
`(4.6 x 150 mm) column, a Spectro Monitor 3200
`UV/VIS_ variable-wavelength detector
`and a
`Merck-Hitachi D-2500 Chromato-Integrator. The
`mobile phase for alprazolam and triazolam con-
`sisted of methanol and water (68:32). The pH of
`the mobile phase was adjusted to 2.7 by addition
`of trifluoroacetic acid. The flow rate was 1.0
`ml/min and the detector was operated at 254 nm.
`For alprazolam, the retention was 2.8 min for the
`ring-open form and 4.7 min for the ring-closed
`form. For triazolam the retention was 2.3 min for
`
`
`
`AQUESTIVE EXHIBIT 1125 Page 0003
`AQUESTIVE EXHIBIT 1125 Page 0003
`
`
`

`

`
`
`
`32
`
`T. Loftsson et al, {International Journal of Pharmaceutics 212 (2001) 29-40
`
`the ring-open form and 3.9 min forthe ring-closed
`form. The mobile phase for midazolam consisted
`of pH 7.2 aqueous 0.004 M phosphate buffer,
`acetonitrile and triethylamine (55:45:0.1). The
`flow rate was 1.5 ml/min and the detector was
`operated at 240 nm. The retention time was 2.6
`min for the ring-open form and 4.2 min for the
`ring-closed form.
`Whenthefraction of ring-open form was deter-
`mined the concentration of the closed form was
`determined right after dissolving the benzodi-
`azepine in the aqueous buffer solution, containing
`either no cyclodextrin or 10% (w/v) cyclodextrin,
`and again 24 h later (ie. after equilibration at
`23°C). Preliminary experiments had shown that
`equilibrium between the closed and the open form
`was attained within 3 h at 23°C and that no
`degradation of cither the ring-open or the ring-
`closed form occurred during the 24-h experiment.
`
`2.4, Kinetic studies in aqueous buffer solutions
`
`A stock solution (1.0 x 10-7 M)of the drug to
`be tested was prepared in a 0.1 M aqueous hydro-
`chloric acid solution (pH 1). This solution was
`equilibrated in a 37°C water bath for 3 h. This
`was to ensure that only the ring-open form was
`present in the stock solution. Cyclodextrin, etha-
`noi or dimethy! sulfoxide (DMSO) was dissolved
`in, or mixed with, pH 7.5 aqueous 0.5 M tris(hy-
`droxymethyl)aminomethane (Tris) buffer solution
`and the solution equilibrated at 37°C. At time
`zero, 30 ul of the stock solution was added to 1.5
`ml of the buffer solution, mixed for a couple of
`seconds on a vortex mixer, and placed again in
`the 37°C water bath. At various time points sam-
`ples were withdrawn from the reaction media and
`injected into a HPLC system (see Section 2,3),
`Both the ring-open and the ring-closed forms
`could be detected by HPLC and the disappear-
`ance of the ring-open form was proportional to
`the appearance of the ring-closed form. Thefirst-
`order rate constants (k,,,) for the disappearance
`of the ring-open form was calculated by linear
`regression of the natural logarithm of the peak
`height versus time plots.
`
`2,5, Kinetic studies in human serum
`
`The rate constant for the ring-closing reaction
`was determined in serum. The previously de-
`scribed (Section 2.4) stock solution of the drug
`(15 pl) was added to 1485 pl of serum which had
`previously been equilibrated at 37°C. After thor-
`ough mixing on a vortex mixer for a couple of
`seconds the solution was placed in a 37°C water
`bath. Sample (100 ul) was withdrawn from the
`solution at various time points and mixed with
`900 pl of ice cold methanol and the solution
`sonicated for 1 min. Then the solution was cen-
`trifuged and the clear supernatant analyzed by
`HPLC,
`
`2.6, Formulation of the aqueous nasal spray
`solution
`
`The phase solubility of midazolam was deter-
`mined in a medium which closely resembled the
`aqueous nasal spray vehicle,
`ie. 7-13% (w/y)
`SBEBCD, 0.10% (w/v) HPMC, 0.02% (w/v} ben-
`zalkonium chloride, 0.10% (w/v) EDTA and
`0.43% (v/v) concentrated phosphoric acid. Excess
`midazolam was added to this medium and the pH
`adjusted to 4.35 with concentrated aqueous
`sodium hydroxide solution, both before and after
`heating in an autoclave (121°C for 40 min). Then
`the samples were allowed to equilibrate for at
`least 4 days at room temperature and analyzed as
`before (Section 2.2). The exact composition of the
`nasal spray was based on this study. The viscosity
`of
`the nasal
`spray was determined with a
`Brookfield viscometer (UK) fitted with a ULA-
`DIN spindle and an UL sample holder with wa-
`ter-circulation jacket (25°C). The osmolarity of
`the nasal spray was measured by the freezing
`point depression method using a Knauer Os-
`raometer Automatic (Netherlands). The buffer ca-
`pacity of the nasal spray was estimated by the
`titration method using an aqueous 0.1 N sodium
`hydroxide solution. The preliminary evaluation of
`the chemical! stability of midazolam in the nasal
`formulation was performed by determining the
`midazolam concentration after successive heating
`cycles in an autoclave (Midmark M7 SpeedClave).
`Each heating cycle consisted of heating to 121°C,
`
`
`
`
`
`AQUESTIVE EXHIBIT 1125 Page 0004
`AQUESTIVE EXHIBIT 1125 Page 0004
`
`
`

`

`
`
`T. Lofisson et al. / International Journal of Pharmaceutics 212 (2001) 29-40
`
`33
`
`maintaining this temperature for 20 min, and
`cooling to room temperature. The midazolam
`concentration was determined after each heating
`cycle. The total number of heating cycles was
`six. Finally the midazolam nasal spray was
`stored at room temperature (22—23°C) and sam-
`ples collected at 0, 3, 4 and 12 months and
`analyzed,
`
`2.7. Evaluation in humans
`
`i.v. administration was compared in each partic-
`ipant and the maximum serum concentration
`(Cyrax) and time to reach C,,., (tmax} determined.
`In each participant the area under the serum—
`time curve from 0 to 6 h (AUC) was calculated
`after both in. and iv. administration using the
`linear
`trapezoidal method, and the absolute
`availability determined from the AUC,,/D,, over
`AUC,,/D,, ratio.
`
`The study was approved by both the ethics
`committee of the National University Hospital
`and the State Committee on Pharmaceuticals in
`Iceland. Six healthy volunteers (two females and
`four men} were
`recruited
`in
`a non-blind,
`crossover study. After obtaining informed con-
`sent and 8-h overnight fast, each participant re-
`ceived either
`intranasal
`(in.)
`ocr
`intravenous
`(i.v.) application of midazolam. The other appli-
`cation was carried out 7 days later. The partici-
`pants
`continued
`to
`fast
`until
`2
`h
`after
`administration of the study formulation. For i.n.
`administration, the participants received 0.06 mg
`of midazolam per kg body weight (D,,,), or 200—
`300 pi, of the aqueous nasal solution (Unit
`Dose closed spray system from Pfeiffer). For i.v.
`administration the participants received 2 mg of
`midazolam (D,,) in an i.v. solution (Dormicum®
`from Hoffmann—La Roche). Blood samples (5
`ml} were collected from an intravenous catheter
`~ at 3, 10, 15, 20, 30, 60, 120, 180, 240 and 360
`min. Samples were centrifuged and serum col-
`lected and kept
`frozen until analyzed by re-
`versed phase HPLC method (performed by
`Medicinsk Laboratorium A/S, Denmark). The
`serum concentration of midazolam after in. and
`
`eek Be) BN,
`
`rrAe \
`
`
`aN Ci
`Ff
`
`Neen
`
`Cl
`
`Alprazolam
`
`Midazolam
`
`Triazolam
`
`Fig. }. The chemical structures of the benzodiazepine bases
`studied.
`
`3. Theoretical background
`
`the benzodiazepine drugs studied, ie. al-
`All
`prazolam, midazolam and triazolam, contained
`2,3-dihydro-1H-1,4-benzodiazepine
`structure
`(Fig. 1}. Alprazolam and triazolam have a 1H-
`1,2,4-triazole ring fused on the 1,2-carbon—ni-
`trogen bond of the diazepine nucleus (ie. a
`triazolo
`[4,3-a][l,4]benzodiazepine
`structure),
`where as midazolam has a imidazole ring fused
`on the 1,2-carbon—nitrogen bond (i.e. an imi-
`dazo [1,5-a][1,4]benzodiazepine structure). Imida-
`zole is relatively basic (pK, 6.9) compared to
`1H-1,2,4-triazole. Thus, in midazolam the prote-
`nated nitrogen in position 2 on the imidazole
`ring (ic. N-2a) has a pK, of 6.15 whereas in
`alprazolam and triazolam the protonated N-2a
`on the triazole rmg has pX, <1.5 (Walser et al.,
`1978).
`In the diazepine nucleus the protonated
`nitrogen in position 4 (ic. N-4) has been esti-
`mated to be about 2.4 (Cho et al., 1983). In
`aqueous solutions the benzodiazepines undergo
`a reversible and pH-dependent ring-opening re-
`action (Fig. 2) (Han et al., 1976, 1977a,b; Cho
`et al., 1983). The pX, of the primary nitrogen
`formed has been estimated to be about 7.0 (Cho
`et al., 1983). There are some indications that the
`ring-opening should be pH-independent (Cho et
`al., 1983)
`in which case the ring-opening rate
`constant (k,) can be described by
`
`ky = Kyo fant
`
`(1)
`
`where Ayoo is the pH-independent rate constant
`and fim,
`is
`the fraction of benzodiazepine
`which is protonated in position N-4. However,
`Eq.
`(1)
`is kinetically equivalent
`to Eq.
`(2)
`
`
`
`
`
`AQUESTIVE EXHIBIT 1125 Page 0005
`AQUESTIVE EXHIBIT 1125 Page 0005
`
`

`

`
`
`34
`
`T. Loftsson et al. / International Journal of Pharmaceutics 212 (2001) 29-40
`
`
`
`X=Narc
`
`ork,
`
`Ho-&
`aa .
`WO
`pKaabout2.4
`ot 7
`
`HyG~MN
`N é
`
`
`
`Fig. 2. The ring-opening reaction of benzodiazepines.
`
`och
`
`\.,
`
`.about7
`emLe
`
`Tabie |
`The apparent equilibrium constant between the closed and open forms of the benzodiazepines:
`— {openro:
`[elosed}rotal
`“4
`where [open}yoa1 is the total concentration of benzodiazepine which is in the ring-open form and {closed},,,.q,
`concentration of benzodiazepine which is in the ring-closed form at 37°C
`
`is
`
`the total
`
`Cyclodextrin
`
`pH
`
`eq
`K,
`
`Alprazolam
`
`Midazolam
`
`Triazolam
`
`No cyclodextrin
`
`10% (w/v) HPBCD
`
`10% (w/v) SBEBCD
`
`.
`
`10% (w/v) RMBCD
`
`1
`2
`4
`I
`2
`4
`1
`2
`4
`1
`2
`4
`
`20
`6
`0.3
`100
`15
`0.3
`100
`25
`0.8
`100
`10
`0.1
`
`50
`3
`0.f
`100
`2
`<Q!
`100
`6
`0.1
`50
`0.8
`0.1
`
`.
`
`20
`2
`<0.1
`20
`2
`<0.1
`50
`3
`<6.1
`20
`0.8
`<0.1
`
`= kKLfHt],
`(2)
`4. Results and discussion
`
`fone =1-_At) @) 4.1. Solubilization
`
`HB+
`Je= [H*14+,+ K,
`The aqueous solubility of benzodiazepines is a
`function of both the ionization of the drug
`molecule and the ring-opening of the diazepine
`ring. The ring-opening of the benzodiazepine ring
`is pH-dependent and fully reversible (Fig. 2). The
`observed equilibrium constant (X,,) between the
`total concentration of the open and closed forms
`is pH-dependent, strongly favoring the closed
`form at pH above 4, but the open form at pH
`below 2 (Table 1). In general, the cyclodextrins
`
`where &,; is the specific acid catalysis rate constant
`for the ring-opening reaction, [H+] is the hydro-
`nium concentration and f, is the fraction of ben-
`zodiazepine which is not protonated in position
`N-4, Comparable equations can be obtained for
`the ring-closing rate constant (k _,). Under nor-
`mal conditions the ring-open forms of alprazo-
`lam, midazolam and triazolam are chemically
`stable in aqueous solutions.
`
`
`
`
`
`AQUESTIVE EXHIBIT 1125 Page 0006
`AQUESTIVE EXHIBIT 1125 Page 0006
`
`
`

`

`
`
`
`
`
`
`
`(mg/ml)
`
`30
`
`
`
`Solubility(mg/ml) a
`
`Triazolam Solubility
`
`f. Loftsson et al. / International Journal of Pharmaceutics 212 (2001) 29-40
`
`35
`
`appear to stabilize the ring-open forms (i.e. OH*
`and OH}3*) resulting in an increased K,, value at
`low pH. The data presented in Fig. 3 are based on
`solubility studies, quantitative determination of
`the total amounts of the ring-open and ring-
`closed benzodiazepine forms, and the observed
`pX, values of the different benzodiazepine forms
`im pure aqueous solution. From Fig. 3, it is possi-
`bie to estimate the contribution of each species
`(Le. different ionization forms of both the ring-
`open and ring-closed forms) to the overall benzo-
`diazepine solubility in aqueous solutions. For
`example,
`it
`is clear
`that
`the monoprotonized
`(BH*) and diprotonized (BH3*)
`ring-closed
`forms, as well as the monoprotonized ring-open
`forms (OH*), have an insignificant effect on the
`overall aqueous solubility of the three benzodi-
`azepines studied. Only when the diprotonized
`ring-open forms (OH2*) emerge do we observe a
`notable increase in aqueous solubility. Further-
`more, it is apparent that the uncharged cyclodex-
`trins (ie. RMBCD and HPBCD)
`interact
`less
`strongly with OH* and OH$+ than with the
`uncharged ring-closed form B, BH*+ or BH3*+
`(Fig.
`3). However,
`the
`negatively
`charged
`SBEBCDinteracts somewhat more strongly than
`the uncharged cyclodextrins with OH3* resulting
`in enhanced solubilization at low pH. The pK,
`values of midazolam are about 2.4 (N-4) and 6.15
`(N-2a) while those of alprazolam and triazolam
`are about 1.5 (N-2a) and 2.4 (N-4). Thus,
`the
`main reason for greater aqueous solubility of
`midazolam with decreasing pH, compared to the
`other two benzodiazepines studied,
`is the early
`appearance of the protonized forms, especially the
`diprotonized OH2* form.
`Cyclodextrins are able to form 1:1 complexes
`with the protonized forms and, thus, they are able
`to solubilize the positively charged ring-open and
`ring-closed forms (Figs. 3 and 4). However, the
`stability constants of these complexes are some-
`what lower than those of comparable uncharged
`species. It is possible to increase the complexation
`efficacy by adding a small amount of a water-sol-
`uble polymer to the aqueous complexation media
`and heating (Loftsson, 1998; Loftsson et al.,
`1999), For midazclam, SBEBCD was the best
`solubilizer of the three cyclodextrins tested and
`
`Solubility
`
`(mg/ml)
`
`Fig. 3. The effects of cyclodextrins, ionization and ring-open-
`ing on the aqueous solubility of benzodiazepines at room
`temperature (22—23°C). The cyclodextrin concentration was
`10% (w/v). f, mol fraction; B, benzodiazepine base (ring-closed
`form); BH*, monoprotonised benzodiazepine; BH3*, dipro-
`tonised benzodiazepine, OH*, monoprotonised ring-open
`form; OH2*, diprotonised ring-open form.
`
`
`
`
`
`AQUESTIVE EXHIBIT 1125 Page 0007
`AQUESTIVE EXHIBIT 1125 Page 0007
`
`
`
`

`

`
`
`36
`
`. T. Lofisson et al. / International Journal of Pharmaceutics 212 (2001) 29-40
`
`thtal
`Solubility(mg/ml) —__oaun
`
`
`
`aowh
`
`2
`
`3
`
`4
`
`pH 5
`
`6
`
`7
`
`8
`
`Fig. 4. The effects of pH and cyclodextrins on the solubility of
`midazolam in aqueous Teorell—Stenhagen buffer system. No
`cyclodextrin present (O); 16% (w/v} HPBCD (A); 10% (w/v)
`SBEBCD (5); 10% (w/v} SBEBCD and 0.10% (w/v) HPMC
`(m).
`
`addition of 0.10% (w/v) hydroxypropy! methylcel-
`lulose (HPMC) and heating in an autoclave at
`121°C for 20-40 min enhanced its solubilizing
`effect (Fig. 4). The value of the stability constant
`of the midazolam/SBEBCD (1:1) complex was
`determined to be 700 M7! at pH 4.8 but 425
`M~' at pH 4.0.
`
`4.2. Kinetic studies
`
`Equilibrium between the ring-open and ring-
`closed forms is reached within a few minutes upon
`dissolution of the benzodiazepine in aqueous me-
`dia. The equilibrium constants are pH-dependent
`favoring the ring-open forms at low pH and the
`ring-closed forms at physiological pH (Tabie 1). It
`is believed that only the ring-closed forms of the
`benzodiazepines
`are pharmacclogically active.
`Thus, it is important to determine how fast the
`ting closes under physiological conditions. The
`half-life of the first-order rate constant was deter-
`mined in aqueous 0.5 M Tris buffer solution at
`pH 7.5 and 37°C. For alprazolam the half-life in
`pure aqueous buffer solution was determined to
`be 5.3 min, 3.9 min for midazolam and 53 min for
`-triazolam, Addition of cyclodextrins
`to the
`aqueous reaction medium increased the half-life
`of the ring-closing reaction (Table 2). This effect
`of cyclodextrins on the half-life is in agreement
`with the observation that cyclodextrins stabilize
`the ring-open forms (i.e. OH* and OH2+). Or-
`
`ganic solvents such as ethanol and dimethylsul-
`foxide reduce the complexation by competing
`with the benzodiazepines for a space in the cy-
`clodextrin cavity and, thus, reducing the effects of
`cyclodextrins. However, when no cyclodextrin
`was present in the reaction medium both ethanol
`and dimethylsulfoxide increased the half-life for
`the ring-closure of alprazolam and midazolam. In
`the case of triazolam the effects were much less
`pronounced,
`In freshly collected human serum the half-life of
`the first-order rate constant for the ring-closing
`reaction was estimated to be less that 2 min for
`both alprazolam and midazolam (in vitro at
`37°C). For triazolam the half-life was somewhat
`higher but still very short. Thus,
`it can be as-
`sumed that
`ring-open forms of the benzodi-
`azepines close very rapidly upon absorption into
`the systemic circulation.
`In the nasal cavity lipophilic molecules will
`compete with the drug molecules for a space in
`the cyclodextrin cavity in much the same way as
`ethanol and DMSO molecules do in our in vitro
`study, Thus, administration of the ring-open form
`of the benzodiazepines in a cyclodextrin-contain-
`ing nasal spray solution should not have any
`effect on their pharmacological effect. That
`is
`beside enhancing aqueous solubility and delivery
`of the drug molecule through the biological mem-
`brane. However, excess cyclodextrin can decrease
`the drug bioavailability in the nasal spray solution
`(Masson et al., 1999). It is therefore important to
`use just enough cyclodextrin to solubilize the drug
`in the aqueous nasal spray solution.
`
`4.3. Formulation of a midazolam nasal spray
`
`The phase solubility of midazolam in the
`aqueous nasal spray vehicle shows that 12.33%
`(w/v) SBEBCD is required to dissolvé=l7-mgvof
`midazolamin:1aml-of:thevehicle (Fig. 5). To
`ensure that no precipitation will be formed during
`storage, a small excess of SBEBCD is needed.
`Thus, the final formulation contained 14% (w/v)
`SBEBCD. The composition of the aqueous nasal
`formulation was as follows: midazolam (1.7% w/
`v), SBEBCD (14% w/v), HPMC (0.1% w/v}, ben-
`zalkonium chloride (0.02% w/v), EDTA (0.1%
`B\e
`7}
`
`aett
`4
`at
`
`
`
`
`
`AQUESTIVE EXHIBIT 1125 Page 0008
`AQUESTIVE EXHIBIT 1125 Page 0008
`
`
`
`

`

`
`
`
`
`
`T. Loftsson et al. / International Journal of Pharmaceutics 212 (2001) 29-46
`
`37
`
`Table 2
`
`The effects of cyclodextrins and organic cosolvents on the half-life for the rate of ring-closure in aqueous 0.5 M Tris buffer solution
`at pH 7.5 and 37.0°C
`
`Cyclodextrin 10% (w/V)
`
`Organic cosolvent® % (v/v)
`
`Half-life ratio”
`
`
`
` Alprazolam® Midazolam* Triazolam®
`
`
`
`No cyclodextrin
`
`HPBCD
`
`1.0
`1.0
`1.0
`No cosolvent
`10
`1.4
`[.]
`10% EtOH
`0.7
`2.1
`1.8
`50% EtOH
`1.6
`1,3
`13
`10% DSMO
`1.0
`1.6
`1,2
`50% DMSO
`2.0
`6.5
`4.2
`No cosolvent
`14
`4.6
`2,3
`10% EtOH
`13
`27
`1.8
`50% EtOH
`1.7
`5.3
`2.9
`10% DSMO
`1.2
`21
`15
`50% DMSO
`2,3
`13.7
`4.2
`No cosolvent
`1.6
`5.4
`2.5
`10% EtOH
`14
`2.8
`2.2
`50% EtOH
`1.8
`8.0
`2.7
`10% DSMO
`1.2
`2.1
`1.4
`50% DMSO
`2.1
`6.0
`5.2
`No cosolvent
`14
`4.2
`2,6
`10% EtOH
`1.3
`2.6
`1.9
`50% EtOH
`1.7
`3.6
`3.3
`10% DSMO
`
`
`
`1.6 2.150% DMSO 12
`
`SBEBCD
`
`RMBCD
`
`* EtOH, absolute ethanol, DMSO, dimethylsulfoxide.
`>The half-life divided by the half-life in pure aqueous buffer solution (i.e. buffer solution containing neither cosolvent nor
`cyclodextrin).
`* The half-life for formation of alprazolam, midazolam and triazolam in aqueous pH 7.5 buffer solution at 37.0°C was determined
`to be 5.3, 3.9 and 53 min,respectively.
`
`The aqueous nasal spray solution was heated in
`sealed containers for up to six successive heating
`cycles. Each heating cycle consisting of heating to
`121°C for 20 min and cooling to room tempera-
`ture. The midazolam concentration in the solution
`
`w/v), concentrated phosphoric acid (0.43% v/v)
`and water (to 100% v/v}. A concentrated aqueous
`sodium hydroxide solution was used to adjust the
`pH to 4.3. The nasal spray was prepared as
`follows. The solid components were weighed into
`a 100-m! volumetric flask. Phosphoric acid and
`most of the water was added and the solution
`stirred until all solid material had dissolved. Then
`the pH was adjusted to 4.35 with a concentrated
`sodium hydroxide solution understirring. Water
`was added to the mark and the aqueoussolution
`heated in a sealed contained in an autoclave
`(121°C for 40 min). After cooling,
`the solution
`was filtered through a sterile 0.45-ym membrane
`filter
`into amber
`glass vials under
`aseptic
`conditions.
`The stability of midazolam in the nasal spray
`upon heating in an autoclave was investigated.
`
`_——
`
`
`
`a8Solubility(mg/ml)
`
`nw&
`
`o
`
`10
`5
`SBE CD conc. (% wi)
`
`15
`
`Fig. 5. The phase solubility of midazolam in the nasal spray
`vehicle at room temperature (22-23°C).
`
`
`
`
`
`AQUESTIVE EXHIBIT 1125 Page 0009
`AQUESTIVE EXHIBIT 1125 Page 0009
`
`
`

`

`
`
`
`
`3oo
`
`T. Loftsson et al. {International Journal of Pharmaceutics 212 (2001) 29-40
`
`te
`
`P=
`
`2=Nw&HA Volumeof0.1MNaOH(ml)
`_=
`
`Serumconcentration(ng/ini}B888
`
`Fig. 6. Titration curve of the midazolam nasal spray. The line
`represents linear fit of data between pH 3.5 and 5.0.
`
`6
`
`ture. Thus, the ring-open form of midazolam has
`adequate chemical stability in the aqueous nasal
`spray solution. The aqueous nasal spray solution
`showed Newtonian flow characteristics and its
`viscosity was determined to be 2.80 + 0.02 mPas.
`The osmolarity of this solution was determined to
`be 541 + 14 mOsm/kg. The buffer capacity of the
`aqueous nasal spray solution was determined
`from linear fit of the titration curve (Fig. 6)
`between pH 3.5 and 5.0. The buffer capacity was
`determined to be 0.016 M. These results show that
`the nasal spray solution is a low viscosity, some-
`what hypertonic solution with adequate buffer
`capacity to maintain constant pH during storage.
`
`4.4. Evaluation in humans
`
`two women and four
`Six healthy volunteers,
`men, were enrolled in this investigation. The mean
`(+58.D.) age and weightof the participants were
`26.0 (45.7) years and 74.5 (+ 10.8) kg, respec-
`tively. The participants only reported mild to
`moderate irritation within the nasal passage and/
`or throat area following administration of 200—
`300 pl (based on body weight) of the nasal spray
`solution. Plots of midazolam serum concentra-
`tion—time curves for in. and iv. administration
`are shown in Fig, 7 and the main pharmacokinetic
`
`Table 3
`Mean (7 = 6) pharmacokinetic parameters and SEM of mida-
`zolam in healthy volunteers following in. administration of a
`0.06 mg/kg dose or iv. administration of a 2 mg fixed dose*
`
`Pharmacokinetic
`parameter
`
`Intranasal
`
`Intravenous
`
`Mean
`
`SEM Mean
`
`SEM
`
`Crnax (ng/ml)
`tnax (hh)
`
`54.3
`0.25
`
`5.0
`0.04
`
`-
`~
`
`-
`-
`
`4 (bh)
`k, (h7')
`AUC/D, 10° (min