Antimicrobial Agents
`and Chemotherapy
`
`Pharmacokinetics of Sparfloxacin in the Serum
`and Vitreous Humor of Rabbits:
`Physicochemical Properties That Regulate
`Penetration of Quinolone Antimicrobials
`
`Weiguo Liu, Qing Feng Liu, Ruth Perkins, George Drusano,
`Arnold Louie, Assumpta Madu, Umar Mian, Martin Mayers and
`Michael H. Miller
`Antimicrob. Agents Chemother. 1998, 42(6):1417.
`
`Updated information and services can be found at:
`http://aac.asm.org/content/42/6/1417
`
`These include:
`
`REFERENCES
`
`This article cites 53 articles, 22 of which can be accessed free at:
`http://aac.asm.org/content/42/6/1417#ref-list-‘l
`
`CONTENT ALERTS
`
`Receive: RSS Feeds, eTOCs, free email alerts (when new articles
`cite this article), more»
`
`Information about commercial reprint orders: http:/Ijournals.asm.org/site/misclreprints.xhtml
`To subscribe to to another ASM Journal go to: http://journals.asm.orglsitelsubscriptionsl
`
`JournalsASMcrg
`
`LUP0099155
`ALCON 2237
`Apotex Inc. v. Alcon Pharmaceuticals, Ltd.
`Case lPR2013-00012
`
`

`

`ANTIMTC‘RFWHAI. AGFNTs ANT) Cnemornnmpv, June 1998, p. 1417—1423
`0066—4804/98/$04.('l0+0
`Copyright (c?) 1998, American Society for Microbiology
`
`Vol. 42, No. 6
`
`Pharmacokinetics of Sparfloxaciri in the Serum and Vitreous Humor
`of Rabbits: Physieochemical Properties That Regulate
`Penetration of Quinolone Antimicrobials
`
`WEIGUO LIU,l ()ING FENG I.IU,1 RU'I‘H PERKINS,l GEORGE I)RUSAN(),1'2 ARNOLD LOUIE,l
`ASSUMP’I‘A MADU,3 UMAR MIAN,3‘4 MAR'I'IN MAYERS?’4 AND MICHAEL H. MILLER”
`
`Divisions oflnfectioits Diseases1 and Clinical I’l-zarma.calogy,2 Departments of Medicine and l’l‘zarmacologv,
`Albany Medical College, Albany, and Department ofOphthalmology, Monte/inre Medical Center,
`University Hospital for the Albert Einstein College of Medicine," and Department
`of Ophthalmology, Bronx Lebanon Medical Center, Albert Einstein
`College ofllledicine,4 Bronx, New York
`
`Received 29 May 'l997/Returned for modification 1'] December '1997/Aecepted 19 March 1998
`
`We have used a recently described animal model to characterize the ocular pharmacokinetics of sparfloxacin
`in vitreous humor of uninfected albino rabbits following systemic administration and direct intraocular injec-
`tion. The relationships of lipophilicity, protein binding, and molecular weight to the penetration and elimina-
`tion of sparfloxacin were compared to those of ciprofloxacin, fleroxacin, and ofloxacin. To determine whether
`elimination was active, elimination rates following direct injection with and without probenecid or heat-killed
`bacteria were compared. Sparfioxacin concentrations were measured in the serum and vitreous humor by a
`biological assay. Protein binding and Iipophilicity were determined, respectively, by ultrafiltration and oil-
`water partitioning. Pharmacokinetic parameters were characterized with RSTRIP, an iterative, nonlinear, weight-
`ed, least-squares-regression program. The relationship between each independent variable and mean quino-
`lone concentration or elimination rate in the vitreous humor was determined by multiple linear regression. The
`mean concentration of sparfioxacin in the vitreous humor was 59.4% i 12.2% of that in serum. Penetration of
`sparfioxacin, ciprofioxacin, fieroxacin, and ofloxacin into, and elimination from, the vitreous humor correlated
`with Iipophilicity (r2 > 0.999). The linear-regression equation describing this relationship was not. improved
`by including the inverse of the square root of the molecular weight and/or the degree of protein binding. Elim-
`ination rates for each quinolone were decreased by the intraocular administration of probenecid. Heat-killed
`Staphylococcus epidermidis decreased the rate of elimination of fleroxacin. Penetration of sparfloxacin into the
`noninflamed vitreous humor was greater than that of any quinolone previously examined. There was an ex-
`cellent correlation between Iipophilicity and vitreous entry or elimination for sparfioxacin as well as cipro-
`fioxacin, fieroxacin, and ofloxacin. There are two modes of quinolone translocation into and out of the vitreous
`humor: dilfusion into the eye and both dilfusion and carrier-media ted elimination out of the vitreous humor.
`
`Bacterial endophthalmitis is a severe and often blinding con-
`dition (2, 22, 48, 52). While the direct injection of antimicro-
`bials into the vitreous humor is known to improve visual out.—
`come, the roles of systemic antibiotics are less well understood
`(7, 21, 48, 52). Systemically administered antimicrobials com—
`monly used in the therapy of endophthalmitis do not penetrate
`into the noninflamed vitreous humor (24, 48, 52). Following cat-
`aract surgery, the intravitreal injection of antimicrobial agents
`in the therapy of endophtliahnitis, which is primarily due to
`Stapl'tylococcits epidermidis, is currently considered the treat.—
`ment of choice [or most patients (24). However, the potential
`role of systemically administered agents that exhibit better
`penetration into the vitreous humor has not been studied.
`Moreover, neither therapy nor prophylaxis of endophthalmitis
`of other causes (e.g., posttraumatic and hematogenous) or
`microbial etiologies (e.g., Streptococcus pncmnoniae, Bacillus
`spp., and Pseudomonas aemginosa) has been well character—
`i zed.
`
`Since accurate pharmacokinetic data have fundamental im-
`plications for outcome studies of animals and humans, we have
`
`v. ("Jot-responding author. Mailing address: Department of Medicine
`Albany Medical College, 47 New Scotland Ave., Albany, NY 12208.
`Phone: (518) 2615343. lit-1.x: (518) 262-6727. E-mail: michacl_miller
`@cegalcwayamcedu.
`
`1417
`
`developed and validated an animal model in which sequential
`vitreous humor samples can be obtained from a small number
`of rabbits. Based upon the comparison of pharmacokinctic
`parameters in single and serially sampled eyes, we have shown
`that. serial sampling does not. alter ocular pharmacokinctic
`parameters. By this approach, the pharmacokinetic parameter
`estimates from as few as three animals give more accurate data
`than it is possible to obtain with more than 20 times this
`number of animals by the approach of combining single datum
`points from ditterent animals (23, 35, 41, 43, 5]). Our method
`provides more—robust. parameter estimates that. permit.
`the
`characterization of ocular phannacokinetics which are difficult
`to address by the older approach (23, 35, 40, 41, 43, 51).
`Studies in our laboratory (23, 41, 43) and by others (16, 39)
`have shown that quinolones penetrate into the noninflamed
`vitreous better than beta—lactams, aminoglycosides, or vanco—
`mycin (5, 31, 34, 36, 38, 59, 60). Based primarily upon these
`penetration data, systemically administered ciprofloxacin has
`been used to treat. pa tienls with bacterial endophthalmitis (32).
`However,
`the activity of ciprolloxacin against ocular patho—
`gens, particularly coagulase-negative staphylococci, is marginal
`and its penetration is poor relative to that of tleroxacin (43) or
`ol'loxacin (51). Sparfloxacin, a recently introduced quinolone
`antimicrobial (14, 54, 58), is more active against staphylococci
`and appears to penetrate into the noninllanied vitreous better
`
`LUP0099156
`
`

`

`1418
`
`LIU ET AL.
`
`ANTIMICRos. AGENTS CllFM(‘)TllFR.
`
`than ciprotloxacin (16, 23, 41). However, the existing pharma—
`cokinetic data are based upon studies which combine single
`samples from different subjects to generate pharmacokinetic
`estimates. This method is unreliable when used to describe
`
`pharmacokinetic data in humans (55).
`The primary goals of the current. study were threefold. We
`wanted to (i) characterize the ocular pharmacokinetics of spar—
`tloxacin, (ii) compare the relationships of protein binding. lipo-
`philicity, and molecular weight (MW) to the vitreous translo-
`cation (entry and elimination) of sparfloxacin with those of
`other quinolones, and (iii) determine if the elimination of
`these drugs was blocked by probeneeid or 11 tat—killed bacteria.
`
`MATERIALS AND METHODS
`
`Animal model. Adult male, New Zealzmd White rabbits (Milbrook Farms.
`Amherst, Mass.) weighing 2 to .3 kg were used. Animals were obtained and cared
`for in accordance with Association for Research in Vision and ()phthalrnology
`guidelines. The care, anesthesia. and vitreous sampling methods were similar to
`those described previously (43). 'lhe animals were anesthetized with an intra-
`muscular dose of diazepam (2.5 mg) and a subcutaneous dose of urethane (1.62
`g/kg of body weight) given approximately 45 min prior to zu'ttibiotic administra-
`tion. Anesthesia was maintained throughout the sampling period, with adminis-
`tration of supplemental
`intramuscular ketamine (10 mgfkg) and xylazine (0.6
`lug/kg) as needed. Following anesthesia. a 24-gaugc angiocathcter was inserted
`into a marginal ear vein to facilitate antibiotic administration and a second
`catheter was inserted into the central artery of the contralateral ear to obtain
`serum samples. A solution of sparfloxaein (obtained from Rhone-Poulene Rorer
`Pharmaceuticals, lnc., (Tollegeville, Pa.) for intravenous injection was prepared
`with 5 ml of a 5% dextrose in water solution and 0.5 ml of lactic acid (pll 3.6)
`and heated by means of a hot tap water bath. After the sparfloxacin was dis-
`solved, another 5 ml of 5% dextrose in water was added to obtain a final
`concentration of 9.5 mgjml. The solution was administered by a rapid (”I-min)
`intravenous infusion (40 mg/kg) through a marginal car vein, followed by a 1-ml
`flush with 0.9% NaCTl. Serial samples (blood and vitreous humor) were taken at
`0.25, 0.5, 1, 2, .3, 4, (i, and h’ h after drug administration as previously described
`(41). For the determination of sparfloxacin ocular phannawhinctics following
`systemic administration, six animals were used. Iior direct injection studies with
`quinolone with and without probenecid, 20 animals in four groups were used.
`Animals in each group received either sparfloxaein. ofloxacin. ciprofloxacin. or
`fieroxacin; one eye received both probenecid and a quinolone, and the other eye
`received quinolone alone. [for direct-injection experiments, solutions of quino-
`lones alone or in combination with probcneeid or heat-liilled bacteria were
`injected into the midvitreous.
`li‘ive additional animals were used in the heat-
`killed-bacterium experiments. Probenecid was dissolved in 1 N Na()ll and ad-
`justed to pH 8.6 prior to injection. Probenccid was diluted in balanced salt to a
`final concentration of 2.86 [Lg/fill. The concentration of ciprofioxacin (Miles
`Pharmaceutical Division, West Haven, Conn), fleroxacin (Roche Pharma—
`ceuticals. Nutlcy. N.J.). ofloxacin (RWG Pharmaceutical Research Institute.
`Raritan, NJ), and sparfloxacin in the direct-injection experiments was 5 ng/ml.
`I [eat-killed S. spiderrrzido (A'I'CCI 155) was prepared with an overnight inoculum
`following three cycles of ccntrifugation zu'td washing with 0.9% saline. Thereaf-
`ter, cells were spectrophotometrically adjusted to a final
`inoculurn of 109 with
`0.9% saline and then heated to 80"(7 for 20 min. ()ne hundred microliters of 109
`heat-killed S. cpidcrmirlis orgau'tisms was injected via a 30-gauge needle into the
`midvitreous cavity of one eye; the contralateral eye received the same volume of
`0.9% saline.
`lior direct-injection experiments, 100 pl of each quinolone was
`injected into the midvitreous as previously described (43). Following the desig-
`nated sampling period, animals were sacrificed with pentobarbital sodium solu-
`tion (125 rug/kg) and bilateral pneumothoraces.
`Antibiotic assays. To determine sparfloxacin concentrations in the senlm and
`vitreous, a well—ditfusion microbiological assay was used. Prior to analysis, all
`samples were stored at —‘l0""(3. Blood samples were allowed to clot and were
`immediately centrifuged at 1.000 2‘in g for 15 min. The test organism was Esche—
`richia (Tali Kl.16. An inoculurn of '107 organisms/ml diluted 1:10 in 3% brain heart
`infusion agar mixed with Mueller-I linton broth (Difco) adjusted to pll 8.0 with
`1 N NaC)H was used. Wells (4-nun-dizuneter) were cut and 1011.1 aliquots of
`serum or vitreous humor were then pipetted into the wells. The agar was incu—
`bated overnight at 37%? in an ambient-air incubator. Zones of inhibition were
`read to the nearest 0.1 mm with a veniier caliper. Sparfloxacin standards were
`prepared by dissolving 100 ug of drug per ml in 1 mmol of Na()ll per liter; this
`solution was then diluted with either rabbit serum (for serum standards, 24, 12,
`8, 4, and 2 rig/ml) or balamced salt solution (for vitreous standards. 12. 6. 3. 1.5.
`0.75, 0.375, and 0.1875 ug/ml). The sensitivity of the biological assay was 1.6 ng.
`'l'he coetficients ofvariation in the biological assay for the high and low standards
`were 4.3 to 7.5% and 0.4 to 3.1%. respectively. with em assay linearity of 0.99.
`There is little or no metabolism of spa rfloxacin with no biologically active me-
`tabolites (11, .30, 45. 50).
`To compare the sensitivity of the biological assay to that of high-pressure
`
`liquid chromatography (HPLC). sparfloxacin concentrations were also measured
`by llPl.(.‘ according to the method of Homer et al. (11). Samples were run at
`‘25"‘(1 in a (7..., 5-j.r.m column (220 by 2.1 mm) packed with Nucleosil. Sample
`preparation was performed by mixing 20 pl of serum with 130 ul of mobile phase
`to acid precipitate proteins zuid by filtering. The mobile phase (75 ’70 acctonitrile—
`25% 0.1 M | 13130.. adjusted to pll 3.82 with concentrated phosphoric acid) was
`delivered to the column at a rate of 0.2 rnl/min with a Hewlett-Packard (Wil-
`mington, Del.) series 1050 pump. Serum samples were prc ared in pooled rabbit
`serum. Vitreous samples could not be assessed by HPLC because of the low
`sensitivity (sparfioxacin does not fiuoresce) of the assay. ( )ne hundred microliters
`
`of sample was injected by a Hewlett-Packard seri
`1050 autosampler and run
`serially through a Hewlett-Packard 1040A UV detector (240- to 280-1un wave-
`lengths) zu‘id a Hewlett-Packard 1046A fluorescence detector (excitation, 2.80
`nm: emission, 445 nm). Data were collected on a | lewlett- Packard (‘bernstatior1.
`Quantitation of the antibiotic concentrations used peak heights. Antibiotic con—
`centrations in the sertlm zuid vitreous following systemic drug administration
`were determined by HPLC (51): concentrations following direct injection were
`determined by the microbiological assay, The coefficients of variation for the
`high and low standards were 1.4 and 2.2%, respectively.
`Protein quanlilaliun and characterization. Protein concentrations in the vit-
`reous humor samples were detcmiined with Coomassie protein assay reagent
`(Pierce, Rockford, 111.). The (.‘oomassie protein assay was performed by placing
`1
`l-Ll of sample, 9 p.1 of distilled water ((11 11.0), and 240 ill of (‘oomassie reagent
`into each well of a 96-well microtiter plate. The plate was read on an E1. 312e
`Biokinetics Reader (BioTek Instruments. Winooski. Vt.) at a filter width of 630
`nm. To prevent overloading of the sodium dodecyl sulfate (Sl)S)-polyacrylamide
`gels, samples were diluted to a final concentration of <4 jig/ml. Albumin stan-
`dards (rabbit albumin; Sigma. St. Louis. M0.) were run at concentrations of 0.5.
`1. 2, 4. 6. 8, and 10 jig/ml.
`Identification and quantitation of proteins in the vitreous humor were per~
`formed by SDS-polyacrylamide gel electrophoresis Mini-Protean 11 cell, model
`with 1000/500 power supply; Bio-Rad. Hercules. Calif.) and densitomctly (model
`60S video dcnsilomctcr; BioImagc. Ann Arbor. Mich.). Minigels were run ac-
`cording to the method of Laemmli (33). We used a, 12% running gel, a 4.5%
`stacking gel, and a 'I'ris (0.15 M)—glycine (1.92 M)—Sl)S (1%) butter. Samples
`were prepared by using 1 “.1 of sample. 4 pl of dH30, and 5 11.1 of sample
`solubilizcr. Eight microliters of szunple was loaded onto the gel. which was run at
`175 V for 40 to 45 min, The gel was stained with (Ioomassie brilliant blue (.1.
`'1'.
`Baker, lnc., Danvers, Mass.) for 30 min and destained with a 5% acetic acid
`solution. Standards included rabbit serum albumin (0.5. 2, and 4 rig/ml), rabbit
`lens protein, zuid rabbit hemoglobin. Rabbit lens protein was obtained by ho-
`mogenizing surgically resected rabbit lenses after the capsules had been re—
`moved. Rabbit hemoglobin was obtained from rabbit erythrocytes that had been
`washed three times in phospl‘tate—bulIered saline (PBS) and lysed in dHZC); cell
`fragments were removed by centrifugation at 8,000 V: 3 (Micro Centrifuge model
`5415C; Brinkmann Instruments lne., Westbury, N.Y.). An MW standard (mid-
`rangc hit: Enprotech, New York, NY.) and lens protein (diltlted 40kt) were also
`run with each gel. Albumin concentrations in vitreous samples zuid sera were
`determined by densitometry.
`Protein binding. The protein binding was determined by ultrafiltrat ion of 4—ml
`standards at several concentrations of sparfloxacin and other quinolones (1.0.
`5.0. 10. zu‘td 20 rig/ml) through Centriflo CF25 (MW eutolI. 25.000) membrane
`cones (Amicon, lnc. Beverly, Mass.) according to the specifications of the man—
`ufacturer. Standard solutions for each quinolone were prepared with rabbit
`serum (Sigma). Briefly. cones were moistened with dHZC), placed into their
`supports. and dried by ccntrifugation at 1.000
`r; for 3 min. Ultraliltration was
`performed at 780
`g for 10 min.
`lt‘ilter binding was determined by comparing
`drug concentrations in ultrafiltrates prepared with PBS with those in spiked PBS.
`Protein binding was adjusted to account for binding to the filter. Concentrations
`of free drtlg in ultratiltrates were dctcnnined by the bioassay described above.
`Lipnphilicity. The lipophilicities of the quinolones were characterized by de~
`termining their partitioning ratios into octanol and PBS by standard methods
`(15). Briefly, solutions containing 10 ug’ml in 0.1 M phosphate bulIer (pH 7.2)
`were agitated with an equal volume of n-octanol at 25C for 48 h and subse-
`quently centrifuged at 1,870 7’ g for phase separation. The concentrations of
`quinolones in the aqueous phase were then detcnnined by the microbiological
`assay. Partition coefllcicnts were expressed as the ratio of the zunount of the
`compound in the rz-octanol phase to that in the aqueous phase.
`Mathematical nmdeling and statistics. Pharrnacokinetic analyses of the plas-
`ma zu'td vitreous humor concentration-time data following systemic administra-
`tion were perfonned with RSTRIP (Micromath Scientific Software. Salt Lake
`(Tity, Utah), an iterative, nonlinear, weighted, least—squares—regression program.
`The most appropriate pharmacokinetic models were determined by using the
`coeflicient of determination and the RSTRIP model selection criterion, which is
`a modified form of the Akaike (1) information criterion. Noncompartmental
`parameters were estimated by the statistical-moment theory.
`listimations for
`each exponential coefiicient and time constant were computed with the standard
`deviations of each estimate. along with its 95% confidence rzmge, which was
`calculated by using both univariate and support»p1ane approximations for the
`bounds of the 95% confidence range. ()ther standard pha rmacokinetic param—
`eters were determined with computer-generated primary coefiicients and stan-
`dard phannacokinetic equations (26. 27). Parzunetcrs were calculated for each
`
`LUP0099157
`
`

`

`VOL. 42. 1998
`
`PHARMACOKINFTICS OF SPARFIDXACIN TN RABBITS
`
`1419
`
` 0.00
`
`1.110
`
`'1 00
`
`151.00
`
`0.05
`
`
`
`<-:.u:
`
`min
`
`' 0.00
`
`2.00
`
`4.00
`“'11-
`Fix“:
`‘
`MG. 1. Mean concentrations of sparflosacin in the serum and vitreous humor of six rabbits following a single intravenous dose (40 mg/kg). The left graph shows
`the data plotted arithmetically. and the right graph shows the data plotted semilogarithmieally.
`
`animal; population pharmacokinetic parameters were then calculated by a stan—
`dard two-step technique (27).
`To determine the relative contribution of MW. protein binding, zmd hydro-
`phobieity (independent variables) to the penetration of quinolones into the
`vitreous humor, we used multiple linear regression (SYS'I'A'I‘,
`livanston, lll.).
`Penetration was expressed as a percentage by dividing the area under the eon-
`eentration—time curve (AUC) from 0 h to infinity in the vitreous by that in the
`serum following a single dose of each quinolone. To determine the relative
`
`iniporttmce of protein binding, levels of penetration were exprt
`ed as both total
`and free fractions; the latter were calculated as percent penetration (free) =
`percent penetration (total) 7- (1% of protein bound). Since protein concentra-
`tions in the vitreous humor are less than 1% of those in serum and since animals
`with any breakdown of the blood-ocular barrier (BOB) were excluded from
`analysis, for these calculations we assumed that there was no binding in the
`vitreous humor. The logaritluns of the mean penetration amd of the mezm free
`penetration were the dependent variables (33, 50) in systemic-administration
`experiments ln direct-injection experiments, the first—order elimination rate
`half-lives were compared with the logarithm of the partition eoeflieient (3). We
`performed univariatc-lu‘rear-regression zu‘ralysis, employing the octau'rol-water
`partition coefficient (the permeability coefficient |[)]), the square root of the MW,
`the fraction of protein bound, amd a hybrid variable (p/VMW) as independent
`variables. (46. 49) The statistical significance of each of the variables was deter-
`mined univariately (6, 30, 46, 57). For the multiplelinear regressions, each of the
`independent variables was allowed to step in (P --'32 0.05) or step out (P
`0.15).
`
`RE S U LTS
`
`Determination of sparfloxacin concentrations in serum and
`vitreous humor. Because of the small sample sizes (5 to 10 pl)
`used when serial samples were obtained from the vitreous hu—
`mor in our ocular pharmacokinetic model (40, 41, 43), very
`sensitive assay methods were required. As a result, we com—
`pared the sensitivities and reproducibilities of results of HPI.C
`and microbiological assays for sparfioxacin using modifications
`of standard assays previously described by others (1]). The
`sensitivities of the microbiological and IIPLC‘. assays were 1.9
`and 25 ng, respectively. Thus, the biological assay was 14—fold
`more sensitive than IIPLC. For the biological assay, the coef—
`ficients of variation for the high and low standards were 1 and
`4.5%, respectively. No metabolites were found in serum sam-
`ples by Hl’l.(.‘.
`Ocular pharmacokineties of sparflexaein. Data from six
`animals with no breakdown of the blood—vitreous barrier, as
`determined by SDS—polyacrylamide gel electrophoresis, were
`analyzed (Fig. I). Results are plotted arithmetically and semi-
`logarithmically to better demonstrate the relative levels of pen-
`etration and terminal elimination slopes, respectively. Model-
`predicted and actually observed drug concentrations in the
`serum and vitreous were similar (Table 1). Both hybrid and
`derived mieroconstants are given in Table 2. Model—dependent
`analysis gave excellent fit with coefficients of determination for
`the serum and Vitreous of 0.999 and 0.997, respectively. The
`AUCs in the vitreous humor and serum were 14.43 and 22.03
`
`mg - h/liter, respectively. Penetration into the vitreous humor
`was 59.4% : 12.2% of that. in the serum. The terminal elim—
`
`ination rate constants in the vitreous humor and serum were
`
`0.28 and 0.24, respectively. The elimination half-life in the
`vitreous humor was 2.99 h, and that in serum was 2.39 h (1’ “2
`0.05). ()n the basis of the coefficient of determination and
`model selection criterion, vitreous humor and serum antibiotic
`concentration-time data following intravenous administration
`were best-fitted to a two-compartment model.
`Correlation between physicochemieal properties and pro-
`tein binding and ocular translocation. The second goal of this
`ocular pharmacokinetic study was to determine the relation-
`ship ol lipophilicity, MW, and protein binding to translocation
`across the blood—ocular barrier of the quinolone antimicrobial
`following systemic and direct. injections. The translocation of
`sparfloxacin was compared to those of three other quinolones
`(ciprolloxacin, Heroxacin, and olloxacin) [or which we have
`previously shown significant (lilIererices in levels of ocular pen—
`etration (Fig. 2) (23, 41, 43, 51). Among the [our quinolones
`studied, levels of penetration differed by an order of magni-
`tude; levels of ciprofioxacin and sparfioxacin penetration were
`5.5 and 59%, respectively. The eifects of three independent
`variables on ocular penetration were considered in the multi-
`ple-linear-regression model:
`lipophilicity, MW, and protein
`binding.
`Table 3 shows the ocular penetration of each drug along
`with its MW, level of protein binding, and partition coellicient.
`Only the lipophilicities were statistically significant when ex—
`amine d univariately. This relationship is described by the equa—
`tion log(mean percent vitrcal penetration) = 2.739(p) + 0.59,
`where p is the octanol—water partition coeflicient (r2 :2: 0.999,
`P i. 0.001). Multiple linear regression was then undertaken
`after considering additional variables, including MW and the
`
`TABLE 1. Comparison of measured and pharmacokinetic—modcl—
`predicted sparfloxaein concentrations in serum and Vitreous
`htunor following systemic administration
`Sparlloxaein level (rig/ml) in:
`Time
`(11)
`Serum
`Vitreous humor
`M easu red
`Predicted
`M ea su red
`Predicted
`
`
`1.6642
`1.638 t 0.844
`12.355
`12.42 i 3.86
`0.25
`2.4495
`2.702 t 0.765
`7.9837
`7.89 i 1.86
`0.5
`2.9741
`2.836 : 0.484
`4.6642
`4.708 : 0.72
`1.0
`2.7177
`2.706 : 0.496
`3.0106
`3.429 : 0.54
`2.0
`2.2234
`2.267 t 0.462
`2.2635
`2.059 t 0.27
`3.0
`1.7928
`1.909 t 0.424
`1.7190
`1.589 t 0.12
`4.0
`1.1605
`1.056 : 0.224
`0.9928
`0.961 : 0.045
`6.0
`
`
`
`
`0.611 1— 0.0858.0 0.7509 0.5734 0.781 r 0.235
`
`LUP0099158
`
`

`

`'1 420
`
`1111 FT A1,.
`
`ANTIMICROR. AGENTS CHEMOTHFR.
`
`
`
` Sample % Penetration L SI) All(.‘. (mg - b/liter)
`
`TABLE 2. Kinetic parameters of sparlloxacin following intravenous administration
`Mean time (11)“ 1 S1) for:
`
`A
`I}
`or.
`B
`Bit.“
`22.31 i 3.70
`'19
`18.85 : 11.97
`5.19 i 1
`3.33 i 1.63
`0.28 i 0.04
`2.39 i 0.29
`Serum
`
`
`—4.80 i 1.54Vitreous 13.06 i— 2.33 4.33 i— l 59 2.01 i 1.57 0.24 i 0.06 2.99 i 0.76 59.38 : 12.26
`
`
`
`
`
`
`
`"A, zero time intercept for or. phase: II, zero time intercept for B phase; or, distribution phase: [3, elimination phase: film, terminal elimination half-life.
`
`free fraction of drug in the serum available for transport; these
`additional variables did not improve model fit.
`To determine if carrier-independent translocation across the
`blood-ocular barrier following direct injection also correlated
`with the physicochemical properties of quinolones, we also de—
`termined the association between lipophilicity and drug elim—
`ination following direct. injection into the vitreous humor in 20
`animals. ()ne eye received the quinolone alone, and the other
`eye received both the quinolone and probenecid. 'l‘he elimina-
`tion half-lives for ciprofloxacin, fleroxacin, ol'loxacin, and spar-
`lloxacin were 4.41, 3.35, 3.04, and 2.78 h, respectively (Table 4).
`As with systemic—injection experiments, there was an excellent
`correlation between lipophilicity and elllux (r2 T) 0.99, P <
`0.01); MW and the free fraction did not improve model fit. The
`relationship is described by the equation lI/zs = (—1.8172)
`(logmp) + 2.1239, where [1/28 is the half-life at beta phase.
`Effects of probenecid and hea t-killed bacteria on quinolone
`elimination following direct injection. Since the renal elimina—
`tion of quinolones and beta—lactam antibiotics in humans and
`rabbits is blocked by probenecid and since the ocular elimina-
`tion of the carrier-mediated export of beta-lactams from the
`vitreous humor is blocked by both probenecid and heat-killed
`bacteria (8, 25, 37), we examined the elIect.s of each on the
`elimination of quinolones following direct. injection. As shown
`in Fig. 3 and Table 4, probenecid significantly increased the
`elimination half-lives of ciprol‘loxacin, fleroxacin, and spar-
`l'loxacin (1’ < 0.05). While probenecid also increased the elim-
`ination half-life of ol'loxacin (4.15 h with probenecid versus
`3.04 h without), this difference was not significant. (P = 0.15).
`Heat—killed bacteria also increased the elimination half—life of
`
`lleroxacin 1.42—fold (P a, 0.01). The clIect.s of inflammation on
`the elimination rates of ciprol‘loxacin, ot'loxacin, and spart‘loxa-
`cin were not tested.
`
`DISCUSSION
`
`Because of the small sample sizes obtained from the vitreous
`humor, very sensitive assay methods were required. When we
`compared the sensitivities of Hl’1.(.‘ and microbiological assays
`for spart‘loxacin, the latter method proved to be 14-fold more
`
`sensitive than IIPLC with coefficients of variation for the high
`and low standards of 1 and 4.5%, respectively. No metabolites
`were found in serum samples by IIPLC. Recent. studies in our
`laboratory with the quinolone ciprolloxacin have shown that, in
`general,
`the sensitivities and reproducibilities of results of
`IIPLC and biological assays are equivalent. However, the ac—
`tivities of quinolones dillcr in the presence and absence of
`mierobiologically active metabolites and quinolones differ in
`their capacities to fluorescc. For compounds with active me—
`tabolites (50) (e.g., ciprolloxacin and olloxacin), IIPLC is the
`preferred assay method when drugs are administered system—
`ically. On the other hand, for compounds like lleroxacin, for
`which ther \ are no active metabolites (57), the biological assay
`is preferred (43). Like lleroxacin, sparlloxacin (lilIers from cip—
`rolloxacin and ofloxacin by not having biologically active me—
`tabolites. However, unlike other quinolones, sparfloxacin does
`not tluoresce; the sensitivity of IIPLC assays with quinolones is
`incr ‘ased by at least. an order of magnitude when fluorescent
`compounds are used. As a result, when doses of sparfloxaein
`that mimic those achieved in the sera of humans were used, the
`IIPLC assay was not sufficiently sensitive to m ‘asure drug
`concentrations in ocular fluid.
`
`Sparlloxacin showed excellent. penetration into the vitreous
`humor, with mean concentrations in the Vitreous humor of
`uninl'lamed eyes of 59.4% : 12.2% of that in the serum.
`Following systemic administration, the elimination half-life
`from the vitreous in rabbits was 3.34 h and that from the serum
`was 2.2 h. The terminal-elimination half-life and maximum
`
`concentration of spart‘loxacin in human serum were 17.6 h and
`1.6 ug/ml, respectively. (30) The maximum concentrations in
`the serum and vitreous of rabbits following a 40-mg/kg bolus.
`achieved at approximately 1 hour after intravenous adminis-
`tration, were 12.43 and 2.84 rig/ml, respectively. While albino
`rabbits were used in this study, previous experiments in our
`laboratory have shown that the levels of penetration of other
`quinolones, namely, ofioxacin and ciprofioxacin, into the vit-
`reous humor are identical
`in pigmented and nonpigmented
`animals (51).
`Recent pharmacokinetic studies by (Tochereau-Massin and
`colleagues with pigmented, uninfected rabbits showed the
`
`100
`
`‘10.
`
`Penetration i
`%Ocular
`
`(1.00
`
`0.10
`
`0.30
`0.20
`Partition Coalllcienl
`
`0.40
`
`0.50
`
`4:.
`
`(43
`
`r0
`
`
`
`Hal!Life[hours]
`
`1
`0.03
`
`\
`
`.|—n_n_.i_|_L_|_‘L_l
`0.1
`1
`Partition Coefficient
`
`li‘lCi. l. (A) Relationship between the partition coefiicients for ciprofloxacin (O), fleroxacin (O), ofloxacin (A), and sparfloxacin (° ) and levels of penetration into
`the vitreous humor.
`(15) Relationship between the partition coefficients for these quinolones (same symbols) and the elimination rate half-lives following direct
`intravitrcal injection.
`
`LUP0099159
`
`

`

`VOL. 42, 1998
`
`PHARMACIOKINFTICS OF SPARFIOXACIN IN RABBITS
`
`142']
`
`TABLE 3. Relationship 01' lipophilicity, protein binding, and MW
`to the penetration of four quinolones into the Vitreous humor
`()cula r
`Pa rtition
`Protein
` ()uinolone penetration (%) eoeflleient binding (’70) MW
`
`
`
`
`('Iiprofioxacin
`5.5
`0.056
`23
`331.3
`l’leroxaein
`14
`0.200
`31
`369.3
`Olloxaein
`30
`0.330
`33
`360.4
`
`Sparfioxacin 392.4 59 0.431 42
`
`
`
`
`maximum vitreal concentration to be 5.6 ug/ml, with a level of
`penetration of 54% following systemic injection of 50 trig/kg
`(16). Those authors also showed that spartloxacin was more
`elllcacious in the therapy of staphylococcal endophthalmitis in
`rabbits than systemically administered vancomycin or amikacin
`(37). lmportantly, newer quinolones such as sparfloxacin (35)
`and ol‘loxacin (51) not only show better penetration into the
`vitreous humor than other quinolones such as ciprolloxacin
`(P <. 0.05) but also are more active against gram—positive bac—
`teria commonly isolated from patients with endophthahnitis.
`Using multiple linear regression, we have shown that there
`was an excellent correlation