`PHARMACEUTICAL
`SCIENCES ®
`
`JULY 1979
`VOLUME 68 NUMBER 7
`
`RESEARCH ARTICLES
`
`5-Azacytidine Hydrolysis Kinetics
`Measured by High-Pressure Liquid Chromatography and
`13C-NMR Spectroscopy
`
`KENNETH K. CHAN X, DONALD D. GIANNINI•, JAMES A. STAROSCIK, and
`WOLFGANG SADEE*
`Received August 18, 1978, from the School of Pharmacy, Stauffer Pharmaceutical Sciences Center, University of Southern California, Los
`Accepted for publication January 22, 1979.
`*Postdoctoral fellow, Chemistry Department, California Institute of
`Angeles, CA 90033.
`Technology. Present address: Eastman Kodak Research Laboratory, Rochester, NY 14650.
`!School of Pharmacy, University of California,
`San Francisco, CA 94143.
`
`Abstract C Hydrolysis of 5-azacytidine, an experimental anticancer
`drug, in aqueous buffers was measured using a high-pressure liquid
`chromatographic (HPLC) procedure and a 13C-NMR method. The for(cid:173)
`mer utilized a 17.5-µm Aminex A-6 strong cation-exchanger column
`eluted with 0.4 M, pH 4.6 ammonium formate buffer at a flow rate of 0.4
`ml/min. The hydrolysis sequence as well as the existence of a labile in(cid:173)
`termediate, N -formylguanylribosylurea, was unequivocally established
`using 6- 13C-5-azacytidine and NMR spectral techniques. A reversible
`ring opening step to the N -formylguanylribosylurea with an equilibrium
`constant of 0.58 ± 0.03 between pH 5.6 and 8.5, followed by an irreversible
`formation of guanylribosylurea, was found by HPLC. The data confirm
`previous assumptions on the hydrolytic kinetics. The pH dependency
`of hydrolysis was examined, and the hydrolysis profile gave a normal V
`shape with the most stable pH at 7 .0. Rather stable intravenous dosage
`forms can be formulated.
`Keypbrases C 5-Azacytidine-hydrolysis kinetics, aqueous solutions,
`high-pressure liquid chromatography, 1=1C-NMR □ Hydrolysis kinet•
`ics-5-azacytidine, aqueous solutions C Antineoplastic agents-5-aza(cid:173)
`cytidine, hydrolysis kinetics, aqueous solutions
`
`A cytidine analog, 5-azacytidine1 (I), first synthesized
`in 1964 (1) and subsequently isolated from Streptoverti(cid:173)
`cillium ladakanus (2, 3), has shown antitumor activity
`against several animal neoplasms including L-1210 leu(cid:173)
`kemia (4) and T-4 Jymphoma (5). Clinically, it has also
`demonstrated activity against various solid tumors (6) as
`well as leukemias (7) artd is currently undergoing phase II
`trials (8).
`This compound has long been known to be unstable in
`aqueous solution. Its }ability has been attributed to the
`
`1 4-Amino-1-8-D-ribofuranosyl-l,3,5-triazin-2-one, NSC-102816, CAS Reg. No.
`320-67-2. Supplied by the Drug Synthesis and Chemistry Branch, Developmental
`Therapeutics Program, Division of Cancer Treatment, National Cancer Institute,
`Bethesda, MD 20014.
`
`facile hydrolytic cleavage across the 5,6-bond (9-11).
`Consequently, proper formulation has been a problem in
`its clinical use. In addition, not only has the metabolic
`evidence indicated that ring cleavage is a major process of
`its disposition {12-14), the facile process may have a pos(cid:173)
`sible relationship to its still unclear overall biological ac(cid:173)
`tivity. Therefore, the nature and sequence of its hydrolysis
`must be understood. While hydrolysis of I has been in(cid:173)
`vestigated (9-11, 15), the lack of a suitable analytical
`method has hampered a complete and unequivocal kinetic
`analysis, although certain information concerning hy(cid:173)
`drolysis has been obtained through spectroscopic resolu(cid:173)
`tion in a complex hydrolytic mixture (10).
`Recently, specific information was obtained through
`TLC, NMR (11), and high-pressure liquid chromatography
`(HPLC) (15), including the isolation and identification of
`the labile intermediate; however, a systematic kinetic
`analysis is still lacking. This paper describes an HPLC
`method for the simultaneous analysis of i and its labile
`intermediate, N-formylribosylguanylurea (II), which oc(cid:173)
`curs during formation of the hydrolytic product 1-
`,B-n-ribofuranosyl-3-guanylurea (III). Corroborated with
`Fourier transform 13C-NMR, using a 6- 13C-5-azacytidine
`previously synthesized in this laboratory (14), a detailed
`kinetic analysis of the hydrolysis of I in aqueous solution
`is presented.
`
`EXPERIMENTAL
`Chemicals and Reagents-All solvents were either analytical or
`liquid chromatographic grade. 5-Azacytidine was greater than 99% pure
`by HPLC and was used without further purification. Ninety percent la(cid:173)
`beled 6- 13C-5-azacytidine was synthesized as described previously (14).
`l-/l-n-Ribofuranosyl-3-guanylurea (III) was synthesized according to
`
`0022-3549/ 7910700-0807$01.00/ 0
`© 1979, American Pharmac8Utlcal Association
`
`Journal of Pharmaceutical Sciences I 807
`Vol. 68, No. 7, July 1979
`
`Apotex v. Cellgene - IPR2023-00512
`Petitioner Apotex Exhibit 1014-0001
`
`
`
`the procedure of Pithova et al. (9). Sodium formate2 and 90% enriched
`1- 13C-sodium acetate3, used as internal standards, were also pure by
`1C-NMH.
`"
`Analytical Procodure-Methanolic or aqueous solutions of 5-aza(cid:173)
`cytidine were injected into a liquid chromatographM via a 20-µI high(cid:173)
`pressure sample injection valve5. The solution was eluted into a 500 X
`2-mm i.d. stainless steel column~ packed with 17.5-µm Aminex A-66
`strong cation-exchanger with 0.40 M, pH 4.6 ammonium formate huff er
`at a flow rate of0.4 ml/min. The components emerging from the column
`were detected via a UV detector"' set at 254 nm. The UV-ahsorbing
`components were quantitated using either peak height or peak area as
`estimated via a mechanical disk integrator on a strip-chart recorder7• No
`~ignificant difference was observed between these two quantitative
`methods.
`13C-NMR- NMH studies of 5-a?.acytidinc, 6- 13C-5-azacytidine, and
`guanylrihosylurea were recorded in dimethyl sulfoxide or in aqueous
`buffers at pH 8.0 via a 25.2-MHz Fourier transform NMR spectrometer.II.
`Typical parameters for the mea~urements were: acquisition time, 0.8 sec;
`pulse delay, 1.2 sec; sweep width, 5000 Hz, 8 K data points; and tip angle,
`:ioo.
`For kinetic measurements, an aqueous solution of 90% enriched 6-
`1'1C-5-azacytidine, 0.1 M in pH 8.0, 0.067 M phosphate buffer, was placed
`in the !!8mple tube, which was inserted into the spectrometer probe. The
`probe temperature was maintained at 38 ± 0.1 °. Rapid pulses at a pulse
`width of 8 were generated, and the appropriate number of transients was
`accumulated to acquire an adequate signal-to-noise ratio. A timer was
`started when the 5-azacytidine solution wa,; mixed and placed inside the
`spectrometer. At the end of each accumulation, the time was noted.
`Kinetic Measurements-The hydrolytic degradation of 5-azacytidine
`wa~ studied quantitatively hy HPLC at 25 and 37° in 0.067 M sodium
`phosphate huffer and acetate buffer at pH 4.5-9.1.
`All experiments were duplicated. A gas-tight syringe was filled with
`a solution of 5-azacytidine at either 4.10 X 10- 4 or 8.20 X 10-4 Min a
`phosphate buffer of the desired pH. For the 25° runs, the filled syringe
`was allowed to remain at room temperature in the high-pressure injector
`valve module. At approximately IO-min intervals, aliquots were pushed
`into the sample loop and injected immediately. For the 37° runs, the
`syringe was filled, capped, and placed in a constant-temperature hath.
`It was then removed at specific intervals to fill the injector loop and
`quickly replaced in the bath.
`The 5-azacytidine concentrations were followed by either the peak
`height or the peak area method. Stability evaluations were made in
`Hingcr's lactate and in normal saline at room temperature as well as at
`refrigerated temperature (5°).
`
`RESULTS
`Drug Assay-With either methanolic or aqueous solutions of 5-aza(cid:173)
`cytidine kept at ice temperature before injection. a straight line was oh(cid:173)
`served in the 0.041-8.20 X 10-4 M range when either peak heights or peak
`areas were plotted against concentrations. The variance of 10 separate
`injections at 0.20 X 10-4 M was less than 1%.
`When 5-azacytidine was dissolved in a phosphate buffer, 0.067 Mat
`pH 8.0, and chromatographed, in addition to a peak emerging at 3.4 min
`(peak I) corresponding t.o 5-azacytidine, a small, slower running peak
`emerged at 7.3 min (peak 2) and the intensity of peak 2 appeared to in(cid:173)
`crease with time (Fig. 1). Their peak height ratio (peak 1 to peak 2) ap(cid:173)
`peared to be constant at approximately 3:2 after 150 min. The intensities
`of these two peaks then diminished with time while maintaining an es(cid:173)
`sentially constant ratio.
`Rach component was collected into a separate tube at room tempera(cid:173)
`ture. Heinjection of peak 1 immediately following collection yielded a
`single peak corresponding to 5-azacytidine. However, similar treatment
`of peak 2 gave two components corresponding to peaks 1 and 2, the latter
`at a higher intensity. After a longer period, repetition of the injection of
`either peak I or 2 revealed two peaks. this time with a peak 1 to peak 2
`ratio of approximately :l:2 but with both at much diminished intensities.
`Collection of peak 2 at ice temperature did not appear to change the re(cid:173)
`versal of peak 2 to peak 1.
`
`2 Mallinckrodt.
`~ Merck & Co., St. Louis, Mo.
`4 Chromatronix model 3510.
`"Spectra-Physics, Santa Clara, Calif.
`6 Packed in this lahoratory with the re!lin supplied hy Bio-Rad Laboratories,
`Richmond, Calif.
`7 V11rian Acrograph, model 20, Varian Associate!\, Palo Alto, Calif.
`"Varian XL- 100, Varian Associates, Palo Alto, Calif.
`
`808 I Journal of Pharmaceutical Sciences
`Vol. 68, No. 7, July 1979
`
`Peak 1
`.93%
`
`82%
`
`75%
`
`70'll,
`
`63%
`
`Peak 2
`
`18%
`
`25%
`
`30%
`
`37%
`
`t \ --
`
`l
`
`18 min
`
`35min
`
`47 min
`
`112 min
`
`5 min
`INJEC-
`TION
`Figure l-High-pre:isure liquid chromatograms of ,5-azacytidine, 8.2
`X 10- 4 Min pH 8.0, 0.067 M potassium pho.~phate al 25°, using 0.4 M,
`pH 4.6 ammonium formate at 0.4 ml/min.
`
`Attempts to isolate peak 2 by collecting it into a tube in a dry ice(cid:173)
`acetone bath followed by lyophilization failed9• Dissolving the residue
`in water and injecting it into the liquid chromatograph did not give sig(cid:173)
`nificant amounts of a UV-absorbing peak (254 nm). TLC analysis of the
`residue using the literature procedure (14) indicated that guanylri(cid:173)
`bosylurea (Ill) was a major component.
`13C-NMR Analysis-The naturally abundant proton-decoupled
`NMR spectra of 5-azacytidine and guanylribosylurea spiked with formate
`in aqueous buffer are shown in Figs. 2 and 3, respectively. The chemical
`shift assignments of 5-aiacytidine under the present conditions were
`referenced from reported values (16). and those for guanylribosylurea
`were assigned by direct comparison with those of I (Table I). Differences
`of the chemical shifts of guanylribosylurea from 5-azacytidine were seen
`in C-2, C-4, and C-1, as expected from the structure.
`When a 5-azacytidine solution in pH 8 phosphate buffer was subjected
`to Fourier transform NMR analysis, new sets of signals not accounted
`for on the basis of the 5-azacytidine and ribosylguanylurea spectra ap(cid:173)
`peared with time. A new signal appeared at 171.7 ppm, subsequently
`identified as formate by signal enhancement with the addition of sodium
`formate. Furthermore, the signal at 157 .5 ppm was particularly broadened
`(Fig. 4).
`The kinetic detections of 6- 13C-5-azacytidine hydrolysis in a buffered
`solution are shown in Fig. 5. A known concentration of 1·1C-sodium acetate
`was added to the buffered solution in an attempt to quantitate the de(cid:173)
`composition kinetics. The signals ratio between the C-6 of 5-azacytidine
`and the C-1 of sodium acetate was essentially constant until after 17 min.
`A small signal appeared at 167.8 ppm and increased with time. After 90
`min, a second signal at 171. 7 ppm appeared. Both new signals increased
`at the expense of the C-6 of 5-azacytidine, as indicated by slowly de(cid:173)
`creasing ratios of the signal intensity of the C-6 of 5-azacytidine to the
`C-1 of acetate. After 270 min, the intensity of the lower field signal ex(cid:173)
`ceeded that of the higher field one. The time course of the ratios between
`the C-6 signal of 5-azacytidine and the C-1 of acetate, an added internal
`standard, is shown in Fig. 6.
`The identity of these new signals was confirmed by off-resonance de(cid:173)
`coupled NMR measurements, which assess the 1~C-H coupling. In this
`experiment, all signals exhibited a doublet except the signal of acetate
`(Fig. 5h ), which indicated the attachment of one proton to each carbon.
`The chemical shift of the signal at 171.7 ppm coincided with that of the
`formate. On the basis of the chemical rationale of hydrolysis, UV ab-
`. sorption property via HPLC, and C-H coupling via the NMR off-reso(cid:173)
`nance decoupling experiments, the signal at 167 .8 ppm was assigned as
`derived from the formyl carbon of the labile intermediate, N -formyl(cid:173)
`guanylribosylurea.
`Kinetic Analysis-5-Azacytidine hydrolysis was carried out in 0.067
`M phosphate buffers at pH 4.5-9.1 and at l5 and 37°. Selected hydmlysis
`profiles as measured by concentrations of 5-azacytidine uer.ms times are
`shown in Fig. 7. Either monophasic or hiphasic profiles were observed
`in all of the pH and temperature studies on semilog plots. At low pH as
`well as in the vicinity of neutral pH, biphasic profiles were observed, with
`
`9 Recently, Beisler (II) was ahle to isolate and characterize N-formylrihosyl(cid:173)
`guanylurea by HPLC using a C-18 reversed-phase column and water as the elu11nt.
`In the present case, the low pH and high salt content during the lyophili7.ation
`procedure may have caused the deeomposition of this lahile intermediate.
`
`Apotex v. Cellgene - IPR2023-00512
`Petitioner Apotex Exhibit 1014-0002
`
`
`
`Table I-Chemical ShiCts • oC5-Azacytidine and
`Guanylribosylurea
`
`Carbon
`
`C-4
`C-6
`C-2
`C-1'
`C-4'
`C-3'
`C-2'
`C-5'
`
`166.4
`157.5
`156.l
`91.8
`84.6
`74.9
`69.6
`61.1
`
`III
`162.2
`
`165.4
`85.5
`83.9
`74.3
`71.2
`62.7
`
`• Measured in dimethyl sulfoxide and expressed in parts per million with tetra(cid:173)
`methylsilane as the internal standard.
`
`the most pronounced being at pH 7 and 8. At high pH, i.e., 9.1, the first
`phase was too rapid to be discernible so that an apparently monophasic
`profile was observed. Similar biphasic profiles of 5-azacytidine were
`observed in Ringer's lactate as well as in normal saline, although the
`terminal phases declined very slowly.
`From this evidence for the existence of an initial equilibrium phase
`between N-formylguanylribosylurea and 5-azacytidine, the biphasic
`behavior of these hydrolysis profiles were attributed to the kinetic scheme
`of Pithova et al. (9) (Path A, Scheme I).
`The appropriate rate constants can be solved by making several as(cid:173)
`sumptions: (a) the initial process is a rapid equilibrium, (b) N-formyl(cid:173)
`guanylribosylurea exists at an approximate steady state throughout the
`hydrolysis, and (c) k- 1 » k2. These assumptions were supported by
`subsequent estimated data, generating a self-consistency. Then k 1 and
`k-1 are solvable by:
`
`Ci
`
`Cl'
`
`Cf
`
`Cl'
`
`ct'
`
`10 ppm
`
`c,
`
`Cl
`
`Figure 3- 13C-NMR spectrum of guanylribosylurea (0.15 M) and so(cid:173)
`dium formate (0.32 M) in deuterium oxide-water at pH 10.2 and 30°.
`
`NH,
`
`NH,
`
`NH,
`
`N---lN
`
`H, l~O
`
`HOCO -
`
`~
`
`••
`
`HN~N
`
`'A
`HCO HJd O
`
`HO OH
`II
`Path A
`
`H,N---lN
`
`k,
`
`H,
`
`HN,,,l..O
`
`- "W
`
`HO OH
`Ill
`
`and (17, 18):
`
`(Eq.1)
`
`HO OH
`
`(Eq. 2)
`
`k2 is solvable by the first-order degradation of the second phase:
`
`(Eq. 3)
`
`(Eq. 4)
`
`The estimated rate constants for 5-azacytidine hydrolysis in aqueous
`buffer solutions at different pH values and at two temperatures are shown
`in Table II. The stability profile at 25 and 37° as expressed by plotting
`log k1 uersus pH is shown in Fig. 8. The profile at both temperatures
`followed a typical V shape, with the most stable pH at 7.0.
`
`DISCUSSION
`
`As was first proposed by Pithova et al. (9), 5-azacytidine in aqueous
`solution undergoes hydrolysis according to Scheme I in strong aqueous
`acid. Glycosidic linkage cleavage to yield 5-azacytosine, 5-azauracil, and
`D-ribose (Path B) was the major degradative pathway. In neutral and
`
`1 O ppm
`
`"
`
`u
`
`CI'
`
`Cf
`
`Figure 2- 13C-NMR spectrum of 0.J.5 M .5-azacytidine at 30"'. The
`soluent was deuterium oxide-water.
`
`NH,
`
`Path K N--l~
`Q__~o
`H
`
`0
`N)__N
`+ l;,;-"o
`H
`+ t>-rihoee
`Scheme 1-Propo.~ed hydrolysis reactions of 5-azacytidine after Pitlwva
`et al. (9). Path A is in aqueous buffers, and Path Bis in .~trong acids.
`
`basic media, a facile ring cleavage to yield an instable N-formylguanyl(cid:173)
`ribosylurea intermediate followed by a loss of formate to form ribosyl(cid:173)
`guanylurea was proposed as the major degradative pathway (Path A).
`Notari and De Young (10) proposed a refined scheme to include the hy(cid:173)
`dration step across the 5,6-double bond as the first step.
`In those studies, UV spectral techniques were used to quantify the
`hydrolysis kinetics of I. Using a TLC and PMR method, lsraili et al. ( 15)
`studied the hydrolysis of 5-azacytidine and found that its degradation
`in aqueous buffers as well as in human plasma under various conditions
`followed first-order kinetics, hut only monoexponential declines were
`reported.
`
`10 ppm
`
`15 7 5 PPm
`
`I
`
`I I
`-LL ... ~.,~illl
`
`Figure 4- 13C-NMR spectrum of 5-azacytidine in pH 8.0 phosphate
`bu/fer ((J.J 5 M) a/ ter 4 hr at 30°.
`
`Journal of Pharmaceutical Sciences I 809
`Vol. 68, No. 7, July 1979
`
`Apotex v. Cellgene - IPR2023-00512
`Petitioner Apotex Exhibit 1014-0003
`
`
`
`Table II-Estimated Rate Constants• for 5-Azaeytldine Hydrolysis in Aqueous Buffer Solution&
`
`pH
`
`4.5
`,,.6
`7.0
`7.4
`8.0
`8.5
`9.1
`
`k1 X 10-3 M,
`min- 1
`7.67 ± o.:n
`6.27 ± 0.37
`3.22 ± 0.02
`4.36 ± 0.04
`9.92 ± 0.03
`24.00 ± 1.10
`62.20 ± :J.80
`
`25°
`h- 1 x 10-3 M,
`min- 1
`
`17.50 ± 0.40
`11.20 ± l.20
`5.73 ± 0.04
`7.95 ± 0.24
`17.20 ± 0.50
`39.20 ± 1.8
`101.0 ± 6.0
`
`Keq
`
`0.44
`0.57
`0.56
`0.55
`0.58
`0.61
`0.61
`
`k2 x 10-3 M,
`min- 1
`
`4.10
`l.30
`0.77
`0.93
`1.70
`
`10.80
`
`k1 X 10-3 M,
`min- 1
`
`15.60 ± 0.40
`11.20 ± 0.60
`7.33 ± 0.00
`12.8 ± 0.50
`35.00 ± 1.80
`
`30°
`k-1 x w-3 M,
`min- 1
`
`32.30 ± 1.40
`19.60 ± 1.50
`13.05 ± 0.00
`22.20 ± 0.50
`60.80 ± 1.70
`
`Keq
`
`0.48
`0.68
`0.56
`0.58
`0.58
`
`" Calculations were hased on the molar concentrations of remaining 5-azacytidine since the N -formyl intermediate pos5el!lleS different extinction coefficients ( 10, 11).
`At those times where initial equilibrium was observed and used for calculation, insignificant amounts of irreversible degradation occurred, consistent with the assumptions
`mode, i.,•., k-, » k 2, and with the experimental observations (13C-NMR). This finding is also consistent with the observations in Ref. 11.
`
`None of the previous studies isolated and characterized the N(cid:173)
`formylguanylrihosylurea intermediate. In a preliminary study (19) in this
`laboratory, the N-formyl intermediate was isolated and characterized
`by 1ac.NMR. Recently, using HPLC and PMR, Seisler ( 11) also isolated
`and characterized the N -formyl intermediate and reported its biological
`activity. However, no detailed kinetic analysis of the hydrolysis of I was
`presented.
`The aims of this study were to use 13C-NMR coupled with specific
`stable isotopic labeling to demonstrate unequivocally the existence of
`the unstable intermediates and the sequence of the degradation and to
`URe an assay method to quantitate the kinetics and hydrolysis of 5-aza(cid:173)
`cytidine as a function of pH and temperature.
`iac.NMR-Because of the high sensitivity and speed in data acqui(cid:173)
`sition, Fourier transform 13C-NMR has been used in studying reaction
`mechanisms (20, 21). ln the present studies, the large differences in
`chemical shifts among the imino carbon of I, the N -formyl carbon of II,
`and the formic acid carbon were utilized for analysis. The C-H coupling
`(i f'., doublet) should provide a positive identification of their location.
`Initially, it was anticipated that NMR studies on the basis of naturally
`abundant C- t:l would provide the necessary evidence for the hydrolysis;
`
`C
`
`d
`
`7 min
`
`31 min
`
`L-->-,.J
`SO min
`
`90min
`
`I
`
`h
`
`'-'-Clll81-
`
`/
`g
`
`\
`
`however, no additional signals at the low field region except the formate
`carbon have been revealed in the hydrolysate. Subsequently, it was found
`that, due to the overlapping signal between the C-6 of N -formyl carbon
`and the C-4 and 5-azacytidine, it was not possible to follow the kinetics
`of the generation of II using naturally abundant 13C-compounds. This
`problem was circumvented by the use of 13C-5-azacytidine labeled at the
`C-6 position synthesized previously (14).
`Use of 90% labeled 6- 13C-5-azacytidine revealed the generation of
`N -formylribosylguanylurea and subsequently formic acid from a solution
`of 5-azacytidine in pH 8.0 phosphate buffer. The chemical shifts and
`multiplicity in off-resonance decoupled 13C-NMR and UV absorption
`characteristics detected from the HPLC studies are all consistent with
`the assigned N -formylguanylribosylurea structure. In addition to the
`off-resonance decoupled experiment, the formate carbon identity comes
`from enhancement of signal intensity with addition of sodium formate.
`On the basis of the timed 13C-NMR studies, the generation sequence of
`N -formylguanylribosylurea and formate was proven unequivocally. Al(cid:173)
`though carhinolamine formation as the first step of hydration across the
`5,6-double bond (10) appeared to he reasonable, it.q detection by 13C(cid:173)
`NMR has not yet been accomplished, possibly because of its short life
`or very low concentration. Thus, this study also demonstrates the use of
`stable isotope labeling and 13C-NMR in the study of reaction mecha(cid:173)
`nisms.
`Since the relaxation mechanism and T 1 of 5-azacytidine and N(cid:173)
`formylguanylribosylurea are likely to he similar, it was thought that by
`selecting a reference atom with a similar chemical shift, the degradation
`kinetics could be followed. Thus, 90% 1- 13C-sodium acetate equimolar
`to 6- 13C-5-azacytidine was added to the solution for the kinetic mea(cid:173)
`surement. Decomposition kinetics were followed by a change in the signal
`ratios between the C-6 of 5-azacytidine and the C-1 of acetate (Fig. 6).
`However, as shown, the ratio remained relatively constant for the initial
`several minutes and declined afterwards. The kinetic profile was not
`consistent with the HPLC data and remained difficult to interpret. The
`difference was perhaps due to in part to errors in Fourier transform NMR
`arising from limitations because of the number of data points available
`in the computer system and from the nonuniformity in numbers of pulses
`for each time spent. Therefore, steady state may not have been achieved
`in the experiments with few puLqes.
`
`O 1.5
`j::
`< a:
`>-1-
`iii z
`~ 1.0
`z
`..J < z
`<.!)
`iii
`
`off resonance
`1367 min
`269 min
`153 min
`Figure 5- 1'1C-NMR .~pectrum of 9(J':;, f'nriched 6- 1:ic.s-azacytidine
`W. J.1R M) in pH IW, IW67 M pota.~sium pho.~phate buffer containing
`:JW; deutl'rium oxidf' and J:IC-sudium acetate (0.20 M) at 27°. Figures
`5a-5h are spectra al selected time.~. For complete number of time point.9,
`""" Fi1r fi.
`
`o.s----,,,~oo--2~00~---,,3~00---,4~00--""s~oo.,....-""e~o""o-
`MINUTES
`Figure 6-Kinetic profiles of 90% enriched 6- 1:ic . .s-azacytidine W.1.58
`M) hydroly.9is in pH 8.0, 0.067 M potassium phri.,phate buffer containinR
`30':; deuterium oxide and 1'1(:..,odium acetate (0.20 M) at 27° usin,:
`•ac.NMR measurement.9. Vertical axis is signal intPnsity ratio., between
`,5-azacytidine and .rndium acetate.
`
`810 I Journal of Pharmaceutical Sciences
`Vol. 68, No. 7, July 1979
`
`Apotex v. Cellgene - IPR2023-00512
`Petitioner Apotex Exhibit 1014-0004
`
`
`
`a
`
`60
`
`180
`120
`MINUTES
`
`240
`
`300
`
`b
`
`120
`
`240
`
`420
`360
`MINUTES
`
`480
`
`600
`
`720
`
`100
`
`60
`
`40
`
`20
`
`10
`
`~
`100
`T
`0
`
`60
`
`40
`
`20
`
`10
`
`X
`z
`0
`j:::
`~ a:
`I-z w
`(.) z
`8
`
`100
`
`20
`
`10 ---3~0---6"!!0--~9~0---!"'120
`MINUTES
`Figure 7-Kin.etic profiles uf 5-awcytidine (8.2 X 10-◄ M) hydrolysi.~
`in potassium phosphate buffer.~ using HPLC measurements. Key: (a),
`pH 5.6 and 2.5°; (b), pH 7.0 and 2,5°; and (c), pH 9.1 and 25°.
`
`Kinetics Studies-Using UV measurement at 242 nm, Pithova et al.
`(9) first observed an initial increase followed by a subsequent decline
`when the extinction coefficient of 5-azacytidine in pH 6.6-7.8 borate
`buffers was plotted against time. Using similar UV measurements at two
`different wavelengths, Notari and De Young (10) observed biphasic
`profiles when 5-azacytidine absorbances in buffer solutions were plotted
`against time; the decrease or increase with time of the initial phase de(cid:173)
`pended on whether the sample was acidified initially or not. The initial
`phase decreased with time using initially acidified sample and increased
`using nonacidified 5-azacytidine.
`In the present study (Fig. 7), in buffers from pH 4.5 to 8.5, biphasic
`declines were observed when 5-azacytidine concentrations were plotted
`against time. At pH 9.1, the profile appeared to be monophasic and
`probably resulted from a very rapid equilibrium so that the first phase
`was not readily discernible. On the basis of the kinetic scheme (Scheme
`I), the individual rate constants were estimated (Table II). The direct
`kinetic measurements showed a rapid equilibrium between 5-azacytidine
`and N -formylribosylguanylurea, followed by a slower degradation of the
`latter to ribosylguanylurea.
`As was evident at room temperature and various pH values, the rate
`constants, k- 1, for the ring closure were several times larger than those
`of the guanylribosylurea formation, k 2, consistent with the assumption
`made. This observation is reasonable in view of the chemical nature of
`the cleavages since the latter step requires the breakage of an amide bond.
`The k- 1 values were also slightly greater than the k2 values, the rate
`constant for the ring opening, yielding an equilibrium constant of 0.58
`± 0.03 between pH 5.6 and 8.5. This value is close to the constant ratio
`of 0.41 between II and I reported in water (pH unknown) ( 11). At 37° the
`equilibrium constants (0.58 ± 0.01) remained essentially unchanged (pH
`
`100
`
`60
`
`40
`
`20
`
`.le
`C, 10
`0
`..J
`
`6
`
`4
`
`2
`
`10
`
`9
`
`8
`
`7
`pH
`Figure 8-The pH stability profile of 5-azacytidine (8.2 X I 0- 4 M) in
`aqueous buffer.~ (pota.~sium phosphate buffer, 0.067 M) at 25° (0) and
`37° (.0.).
`
`6
`
`6
`
`4
`
`5.6-8.0). The equilibrium constant decreased slightly at a lower pH (4.5).
`These observations are qualitatively consistent with the HPLC and PMR
`observations as well as with reported data (10, 11).
`pH-Dependent Hydrolysie-As shown in Table II, the rate constants
`of the hydrolysis varied with pH. When log k values were plotted against
`pH at 25 and 37°, a V-shaped curve resulted with both minima at pH 7.0
`(Fig. 8), a value slightly different than the pH 6.5 reported (10). The
`slopes on the basic and acidic portions of the curve were +0.630 and
`-0.152, respectively. A similar trend was also observed if log k2 values
`were plotted against pH. The hydrolysis, therefore, appears to be acid
`and base catalyzed, similar to the other known examples (22, 23). The
`deviation of the slopes from + 1 and -1 on the basic and acidic pH por(cid:173)
`tions of the curve, respectively, suggests that the hydrolysis may be
`susceptible to general acid and base catalyses (17). These pH profiles of
`the hydrolyses were markedly different from those reported by Notari
`arid De Young (10) who described a more complex profile not readily
`interpretable on the basis of acid and base catalyses.
`The hydrolysis kinetics observed were qualitatively similar to those
`reported using·HPLC, PMR, and UV; however, quantitation differences
`were observed with those reported and may be attributed to differences
`in experimental conditions as well as methodologies. Under comparable
`experimental conditions such as those in Ringer's lactate, comparable
`results were obtained.
`Stability Evaluation of 5-Azacytidine in Aqueous Buffers and
`Other Solutions-Although 5-azacytidine hydrolysis is subject to acid
`and base catalyses, the rate constants for the reversible formation of
`N-formlyribosylguanylurea as well as for the irreversible formation of
`ribosylguanylurea are smaller in magnitude than at the basic pH. Since
`ribosylguanylurea itself does not seem to possess significant cytotoxicity,
`rapid degradation of 5-azacytidine to this compound will result in a loss
`of efficacy. Recent biological data suggest that N -formylribosylgua(cid:173)
`nylurea also has little cytotoxicity (11), although the evidence was not
`clearcut. However, other researchers found that aqueous solutions of
`5-azacytidine, after long standing at room temperature, resulted in higher
`cytotoxicity (24, 25). Although formation of N-formylribosylguanylurea
`is reversible, in the absence of more definitive data it is desirable to for(cid:173)
`mulate 5-azacytidine in media where the ring opening reaction is re(cid:173)
`tarded. The pH profile of hydrolysis and kinetic data in Table II show
`that the rate constants for the hydrolysis in acidic pH are smaller in
`magnitude than those in basic pH. Therefore, a pH lower than 7.0 will
`result in slower ring hydrolysis.
`In patients, slow infusion of 5-azacytidine in Ringer's lactate has
`lowered GI toxicity. The present study demonstrated that 5-azacytidine
`is very stable in this medium (pH 6.4), yielding a terminal t 112 of 4.8 days
`at room temperature, a value close to the reported one of 4.2 days ( 15).
`The t 112 in this medium at refrigerated temperature (0-5°) was consid(cid:173)
`erably longer (31.3 days).
`
`Journal of Pharmaceutical Sciences I 811
`Vol. 68, No. 7, July 1979
`
`Apotex v. Cellgene - IPR2023-00512
`Petitioner Apotex Exhibit 1014-0005
`
`
`
`REFERENCES
`
`( 1) A. Piskala and F. Sorm, Collect. Czech. Chem. Commun., 29, 2060
`(1964).
`(2) L. J. Hanka, J. S. Evans, D.S. Mason, and A. Dietz, Actimicrob.
`A/,/ents Chemother., 1966, 619.
`(:\) M. E. Bergy and R.R. Herr, ibid., 1966, 625.
`(4) J. S. Evans and L. ,J. Hanka, Experientia, U, 922 (1968).
`(5) L. J. Hanka and J. F. Clark, Proc. Am. Cancer Res., 16, 113
`(1975).
`(6) A. J. Weiss, J. E. Stambaugh, M. J. Mastrangelo, J. F. Laucius,
`and R. E. Bellet, Cancer Chemother. Rep., 56,413 (1972).
`(7) K. B. McCredie. G. P. Bodey, M.A. Burgess, J. U. Gutterman,
`V. Rodriguez, M. P. Sullivan, and E. J. Freireich, ibid., 57,319 (1973).
`(8) A. ,J. Weiss, G. E. Metter, T. F. Nealon, J.P. Keanan, G. Ramirez,
`A. Swiminathan, W. S. Fletcher, S. E. Moss, and R. W. Manthei, Cancer
`Treatment Rep., 61, 55 (1977).
`(9) P. Pithova, A. Piskala, J. Pitha, and F. Sorm, Collect. Czech.
`Chem. Commun., 30, 2801 (1965).
`(JO) R. K Notari and J. L. De Young, J. Pharm. Sci., 64, 1148
`(1975).
`(l J) ,I. A. Beisler, J. Med. Chem., 21,205 (1978).
`(12) M. Surovcik, K. Raska, ,Jr., Z. Sormova, and F. Sorm., Collect.
`Czech. Chem. Commun., 30, 3370 (1965).
`( 13) G. L. Neil, T. E. Moxley, S. L. Kuentzel, R. C. Manak, and L. J.
`Hanka, Cancer Che mother. Rep., 59,459 ( 1975).
`(14) K. K. Chan, J. A. Staroscik,and W. Sadee,J. Med. Chem., 20,598
`(1977).
`(15) Z. H. lsraili, W. R. Vogler, E. S. Mingioli, J. L. Pirkle, R. W.
`Smithwick, and J. H. Goldstein, Cancer Res., 36, 1453 (1976).
`( 16) A. J. ,Jones, D. M. Grant, M. W. Winkley, and R. K. Robins, J.
`Phys. Chem., 74, 2684 (1970).
`
`(17) A. N. Martin, J. Swarhrick, and A. Cammarata, "Physical Phar(cid:173)
`macy," 2nd ed., Lea & Febiger, Philadelphia, Pa., 1969, pp. 383-387.
`(18) A. A. Frost and R. G. Pearson, "Kinetics and Mechanism," 2nd
`ed., Wiley, New York, N.Y., 1961, pp. 143-196.
`(19) K. K. Chan, J. Staroscik, 0. D. Giannini, and W. Sadee, "Ab(cid:173)
`stracts," APhA Academy of Pharmaceutical Sciences, Atlanta meeting,
`Nov. 1975.
`(20) E. Hadaya and M. E. Kent, J. Am. Chem. Soc., 93, 3283
`(1971).
`(21) J.B. Stothers, I. S. Y. Wang, D. Ouchi.and E.W. Wamhoff, ibid.,
`93, 6702 (1971).
`(22) S. Siegel, L. Lechman, and L. Malspeis, J. Am. Pharm. Assoc.,
`Sci. Ed., 48,431 (1959).
`(23) K. T. Koshy and J. L. Lach, J. Pharm. Sci., 50, I 13 (1961).
`(24) P. L. Lomen, L. H. Baker, G. L. Neil, and M. K. Samson, Cancer
`Chemvther. Rep., 59, 1123 (1975).
`(25) H. H. Lloyd, E. A. Dulmadge, and L. J. Wilkoff, ibid., 56, 585
`(1972).
`
`ACKNOWLEDGMENTS
`
`Presented in part at the Pharmaceutical Analysis and Control Section,
`APM Academy of Pharmaceutical Sciences,