Thermal sterilization of heat‐sensitive products using high‐temperature short‐time sterilization

Thermal Sterilization of Heat-Sensitive Products using High-Temperature Short-Time Sterilization ANGELIKA MANN,1,2 MARKUS KIEFER,2 HANS LEUENBERGER1 1

Department of Pharmacy, University of Basel, Klingelbergstrasse 50, CH-4056 Basel, Switzerland

2

Novartis Pharma AG, CH-4002 Basel, Switzerland

Received 22 September 1999; revised 18 May 2000; accepted 22 September 2000

ABSTRACT: High-temperature short-time (HTST) sterilization with a continuous-¯ow sterilizer, developed for this study, was evaluated. The evaluation was performed with respect to (a) the chemical degradation of two heat-sensitive drugs in HTST range (140± 160 C) and (b) the microbiological effect of HTST sterilization. Degradation kinetics of two heat-sensitive drugs showed that a high peak temperature sterilization process resulted in less chemical degradation for the same microbiological effect than a low peak temperature process. Both drugs investigated could be sterilized with acceptable degradation at HTST conditions. For the evaluation of the microbiological effect, Bacillus stearothermophilus ATCC 7953 spores were used as indicator bacteria. Indicator spore kinetics (DT, z value, k, and Ea), were determined in the HTST range. A comparison between the Bigelow model (z value concept) and the Arrhenius model, used to describe the temperature coef®cient of the microbial inactivation, demonstrated that the Bigelow model is more accurate in prediction of DT values in the HTST range. The temperature coef®cient decreased with increasing temperature. The in¯uence of Ca2 ‡ ions and pH value on the heat resistance of the indicator spores, which is known under typical sterilization conditions, did not change under HTST conditions. ß 2001 Wiley-Liss, Inc. and the American Pharmaceutical Association J Pharm Sci 90:275±287, 2001 Keywords: HTST sterilization; Bacillus stearothermophilus; degradation kinetics;

z value; Bigelow model; Arrhenius model; laminar ¯ow

INTRODUCTION A sterilization process for parenteral products has to meet two requirements; that is, be an effective process with regard to microbial inactivation (sterility) and avoid drug degradation. The degradation kinetics of drug substance in solution may be expressed using the Arrhenius equation: ÿEa

k ˆ A
e RT

…1†

Correspondence to: H. Leuenberger (Telephone: 41-612671500; Fax: 41-61-2671516; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 90, 275±287 (2001) ß 2001 Wiley-Liss, Inc. and the American Pharmaceutical Association

where k is the degradation rate constant, A is the frequency factor, Ea is the activation energy, R is the universal gas constant, and T is the absolute temperature.1 The effectiveness of a sterilization process can be quanti®ed by its microbial inactivation capability, which is measured by the F value achieved in this process. The F value is the sterilizing time in minutes that is equivalent to exposure to a saturated steam environment of 121 C. It is an integrated value derived from the formula: Fˆ

… t1 t0

10

T…t† ÿ 121:1 z

dt

…2†

In eq. 2, t0 is the initial process time, t1 is the ®nal process time, and T…t† is the time-dependent

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

275

276

MANN, KIEFER, AND LEUENBERGER

temperature variable.2 The z value is the temperature difference, which causes a 10-fold change in the D value, which is the decimal reduction time of a population of microorganisms under speci®ed conditions. The same F value can be achieved in sterilization processes using either a high temperature and a short exposure time or a low temperature and a long exposure time. Moist heat sterilization is usually the method of choice, but a large number of pharmaceutics are known to degrade3,4 under conditions encountered during typical moist heat sterilization in an autoclave (121 C, 15 min).6 The preservation of a heat-sensitive drug and destruction of microorganisms during sterilization processes is based on the difference in the rates of degradation and inactivation, respectively. The rate of destruction and inactivation vary with temperature. For a sterilization process to be successful, the microorganism inactivation rate must be much larger than the degradation rate of a heat-sensitive drug under the process conditions. Then, the microorganisms will suffer larger destruction by increased process temperature than the heatsensitive drug. The activation energies associated with chemical degradation (50±150 kJ/mol) are generally much smaller than those associated with the inactivation of microorganisms (250±350 kJ/mol).5 According to the Arrhenius equation, the rate constant of chemical degradation is less in¯uenced by increased temperature than the rate constant of the microbial inactivation because of the different activation energies. Thus, sterilization processes using a high peak temperature in high-temperature short-time (HTST) range (140±155 C) and a short exposure time could make it possible to sterilize heatsensitive drugs with acceptable degradation and the same microbiological effectiveness (F value) as a sterilization cycle at typical conditions (121 C, 15 min).6 To predict the positive effect of an HTST process on the preservation of a heat-sensitive drug, the degradation kinetics and the inactivation kinetics of microorganisms in this temperature range must be known. The drug degradation mechanism must be consistent over the whole temperature range. To evaluate the lethal effect of a new sterilization process as a function of F value, determination of the DT and the z values in this temperature range is needed. For HTST process design, using the F concept, one must con®rm whether the z JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

value determined below 130 C can be extrapolated in the temperature range 140±155 C. For the thermal inactivation of the microorganisms, two models can be used: the Arrhenius model7 (see eq. 1) or the Bigelow model (z value concept).8 In the Bigelow model method, the logarithm of the DT value is plotted versus the corresponding temperature and the reciprocal of the slope of the least square regression line is taken as the z value. The Arrhenius graph is constructed by creating a semilogarithmic plot of the reaction rate constant versus the reciprocal of the absolute temperature. The activation energy is then calculated using the equation Ea ˆ ÿ slope
2.303
R. A frequent argument is made that the Arrhenius model is superior to the Bigelow model, which is the temperature coef®cient model commonly used, because the z value is temperature dependent whereas the activation energy is not. In this study we will compare the Bigelow and the Arrhenius models with regard to the extrapolation of the DT values and reaction rate constants (k) determined in the typical sterilization range (121 C±130 C) to the HTST sterilization range. In the HTST process, the lethality during heating and cooling is of the same order of magnitude as during the isothermal holding interval. A 12log reduction of Bacillus stearothermophilus spores at 140 C is performed within 12s. Precise determination of DT at temperature > 130 C can not be performed with standard methods because of the lag time for heating and the very short time of the DT value in the HTST range (D130 ˆ 9 s). It takes 8 s to heat a capillary tube that is commonly used in heat-treatment experiments up to 130 C.9 The HTST systems currently available, such as Kersys,10 Sterilab,11 and Thermalizer,12 are inadequate to achieve the time±temperature relationship that would permit the determination of DT value in the HTST range. The resulting microbial inactivation during the heating and cooling times does not allow precise determination of the DT values in the high-temperature range. Other systems, Thermoresistometer13 and Piston,14 use direct heating, where the spore suspension on ®lters or paper strips is submitted directly to steam. Spore resistance data determined in these systems can vary signi®cantly from those determined in the product solution. At present, an apparatus for determination of the DT value in product formulations at HTST conditions is not available on the market. The purpose of this study was to develop a continuous-¯ow sterilizer adequate to ful®ll the

HIGH-TEMPERATURE SHORT-TIME STERILIZATION OF HEAT-SENSITIVE PRODUCTS

requirements for kinetics measurements in the HTST range; such as short heating time, ¯exible exposure time, and short cooling time. Therefore, we constructed a countercurrent heat exchanger, where the heating time was reduced to 18 ms by the use of very small tubing diameters (0.12 mm) in the heating zone and a large temperature gradient in the countercurrent heat exchanger between the heating medium and the ¯uid to be investigated. Another requirement was that the microbial inactivation kinetic should be determined in the product solution (indirect heating) and not by direct exposure to steam (direct heating). Both, the chemical degradation of two heat-sensitive drugs and the kinetics (DT, z, k, Ea) of indicator spores were investigated in the HTST range with the continuous-¯ow sterilizer. We compared the two models (Arrhenius and Bigelow) for prediction of temperature coef®cients of microbial inactivation in the HTST range. The in¯uence of pharmaceutical formulations, like concentrations of bivalent cations and pH value < 7, on spore heat resistance was investigated in the HTST range.

EXPERIMENTAL SECTION Materials Tetramethylammonium hydroxide pentahydrate (TMAH) was supplied from Sigma Chemical Company (St. Louis, MO). Potassium dihydrogen phosphate (KH2PO4), dansyl chloride, methylamine hydrochloride, boric acid, sodium hydroxide, and g-cyclodextrin were obtained from Fluka Chemie (Buchs, Switzerland). Acetonitrile (HPLC grade), o-phosphoric acid, disodium hydrogen phosphate, calcium chloride (CaCl2), manganese chloride (MnCl2), and yeast extract were purchased by Merck (Darmstadt, Germany). The drug substances, Octreotide and SDZ EAA 494, were supplied from Novartis Pharma AG (Basel, Switzerland). Trypticase soy agar (TSA) was supplied from BiomeÂrieux (Marcy l'Etoile, France), Nutrient Broth from Becton Dickinson (Heidelberg, Germany), Bacto Agar from Difco Laboratories (Detroit, MI), and trypticase soy broth (TSB) from Oxoid Ltd. (Hampshire, England). All compounds and solvents used were of analytical reagent grade. Distilled water was used for all analyses, and WFI (water for injection)6 was used for microbiological analysis and preparation of spore suspensions.

277

Figure 1. Schematic diagram of the continuous-flow device: (1) sample stirring and cooling, (2) HPLC pump; (3, 5, 6) heating/cooling unit, (4) heating zone; tubular heat exchanger, (7) cooling zone, tubular heat exchanger, (8) sample collection in sterile tube; (FI) flowmeter; (PI) pressure gauge; (T3 , T5 , T6 ) temperature control and measurement; (T41 ; T71 ) temperature measurement at heat exchanger inlet; and (T42 , T72 ) temperature measurement at heat exchanger outlet.

Methods Heat Inactivation in the Continuous-Flow Device The study of spore heat inactivation kinetics at high temperatures requires very short heating and cooling times to prevent spore inactivation during these phases. A continuous-¯ow system (Figure 1) was developed to study spore inactivation kinetics and chemical degradation over the entire temperature ranges of 121 to 155 C and 80 to 160 C, respectively. The ¯uid investigated was pumped continuously through a countercurrent tubular heat exchanger, passing a heating zone, a holding zone, and a cooling zone. To reduce heating time, the temperature in the heating zone was slightly higher than the temperature in the holding zone. The ¯uid leaving the heating zone had the selected temperature of the holding zone. In the holding zone, the target temperature was maintained and the holding time could be varied by different ¯ow rates and different internal diameters of the capillary tube (minimum exposure time was 0.04 s). In the cooling zone, the ¯uid was cooled down rapidly to 5 C. The temperature of the heating media at the holding zone, at the inlet and outlet of heating, and at the cooling zone (T41 , T42 , T71 , and T72 , respectively) was measured with a 100 O platinum resistance temperature detector (Pt 100). Temperature control (Pt 100) and overheating limits are located in each of the three heating and cooling units (3, 5, and 6). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

278

MANN, KIEFER, AND LEUENBERGER

Before and after the heating experiments, we determined if the target temperature in the ¯uid leaving the heating zone was achieved under steady-state conditions by inserting a 0.1-cm NiCr/Ni-type thermocouple directly into the ¯uid stream. All thermocouples and Pt 100 were calibrated before the study with reference temperature baths. The accuracy of the thermometers was  0.5 C. All temperatures measured during a process were transmitted to a printer. The backpressure to prevent boiling of the ¯uid (6.18 bar at 160 C15) was maintained by the pressure drop in the capillary in the cooling zone and by a pressure valve at the outlet of the system. Pressure was controlled with the pressure gauge. The tubular system sometimes became progressively fouled by a buildup of material inside the tubing at high temperature processing. This fouling was indicated by an increased pressure under a constant ¯ow rate. The tubes were replaced when this happened. Samples were taken after the system had reached steady-state conditions and the tubings were rinsed with the investigation medium for at least three times the total residence time in the system. The samples were collected in sterile tubes and kept in the refrigerator until further use. The continuous-¯ow sterilizer was constructed using 1.59-mm o.d., 0.12±1.0-mm i.d. stainless steel tubings. HPLC-type stainless steel compression ®ttings (1/16 inch) and HPLC-type valves (Swagelok, Niederrohrdorf, Switzerland) were used throughout the system. In our continuous-¯ow sterilizer, the heating time was reduced to 18 ms. This time is less than that in any other system and enabled spore kinetics determination up to 155 C. The DT values determined in the continuous-¯ow sterilizer at 121 C correlated well with the DT values determined with the standard ampule glycerin bath immersion method. The pressure buildup in the tubing system did not in¯uence the viable spore count, as determined previously. Determination of Residence Time in the Continuous Sterilizer The Reynolds number in the heating and the cooling section was between 7200 and 9400, indicating the transition range between streamline (laminar) and the turbulent (plug) ¯ow. The Reynolds number in the holding tube was JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

between 1130 and 2300, indicating laminar ¯ow. Streamline ¯ow in the holding tube complicates the calculation of spores surviving the process because of the variable residence time associated with the parabolic velocity pro®le. For a given streamline ¯ow, the rate of total surviving spores emerging from the tube can be calculated using eq. 3. Under turbulent ¯ow conditions, the rate of surviving spores can be calculated by eq. 6. N ˆ N0

… Rw 0

10



ÿDR T


2

R
U
dR

…3†

with L U

…4†

ÿ

p
R2W ÿ R2 4

L

…5†

R ˆ with Uˆ with

R

N ˆ N0
10 DT

…6†

with R ˆ

w L

…7†

where N0 is the initial number of spores/mL, N is the spores/mL surviving a given streamline after passing through a hold tube, RW is the hold tube radius, R is the distance along the radius from the center, R is the residence time of a streamline in the hold tube, U is the velocity pro®le associated with a streamline ¯ow, L is the length of the hold tube,
p is the pressure drop in the hold tube,  is the viscosity of the ¯uid, and w is the mean velocity of the ¯uid (m/s). Using eqs. 3 and 6, the inactivation factors (k, DT value) determined for a parabolic velocity pro®le and the plug velocity pro®le were compared. The reaction rate constants (k) determined using the ideal laminar ¯ow model were larger, by a factor of 1.5, than those determined using the turbulent ¯ow model. The corresponding DT values (k ˆ 2:303=DT ) and the resulting sterilization times required for a 12-log reduction were thus smaller when the laminar ¯ow model was used instead of the plug ¯ow model. The rate

HIGH-TEMPERATURE SHORT-TIME STERILIZATION OF HEAT-SENSITIVE PRODUCTS

constants determined using the laminar ¯ow model were consistently larger by a factor 1.5 over the whole temperature range investigated. Using the rate constants (k), the temperature coef®cients (Ea ) over the whole temperature range were calculated. The temperature coef®cients for the rate constants based on the laminar ¯ow model and the temperature coef®cient based on the turbulent ¯ow model were identical. This result means the use of the two different models for residence time calculation did not in¯uence the temperature dependence of the microbial inactivation reaction. It was concluded that if the plug ¯ow idealization is inappropriately used to calculate the microbial inactivation in an ideal laminar holding tube, the rate constants are smaller by a factor of 1.5 than the real rate constants; however, the resulting DT values and sterilization times are larger than those determined using the appropriate laminar ¯ow model. This result means if the HTST process design is based on these calculations, there is no risk with regard to the process safety. The inappropriate use of plug ¯ow idealization results in the calculation of longer sterilization times. The lethal effect against the microorganisms will then be larger than indicated by the inactivation kinetic parameters. The temperature coef®cients were identical in both models, and thus the two different models do not affect extrapolation of rate constants to higher temperatures. To facilitate the calculation of the inactivation rates, the turbulent ¯ow model was used in this study. Because the temperature coef®cients are not in¯uenced by the two different models and the determination of the sterilization parameters and the sterilization process were performed in the same system, the inappropriate use of the turbulent ¯ow model is only of theoretical interest. To transfer the heat inactivation parameter to another sterilizer system, one has to consider that the reaction rate constants determined in this system are smaller by factor of 1.5 than the true reaction rate constants. As far as the chemical degradation of the drug substance is concerned, no difference in reaction rates is to be expected if the plug ¯ow idealization is used instead of the streamline ¯ow model. Ramayya et al.16 have shown that for small values (< 20%) of conversion, the systematic error in k evaluated by misusing the plug ¯ow model approximation to treat ideal, laminar reactor data, does not exceed 10% of its true value. On the other hand, for large values of conversion (for

279

example the microbial inactivation reaction), the systematic error can be as large as 20% or more. In our experiments, the conversion rate for the chemical degradation was in the range of 0.5 to 10%. For this conversion range, a systemic error in k was < 5%. The Ea for a ®rst-order reaction obtained by plug ¯ow treatment of the data taken from an ideal laminar ¯ow does not exceed in 10%. Thus, in the worst case, the systematic error in k and Ea due to a misuse of the plug ¯ow idealization is comparable in magnitude to random errors introduced into k and Ea from uncertainties in analytic techniques. Determination of Octreotide and Its Degradation Products A 0.5-mg/mL solution of Octreotide adjusted to pH 4.2 with o-phosphoric acid was used for kinetic measurements. The concentration of Octreotide before and after thermal stress was quantitated by a reversed-phase high performance liquid chromatography (HPLC) method. The HPLC system (consisting of Spectra Physics system Compaq 4/66, pump P4000, autosampler AS3500, and Focus UV-Detector; Spectra Physics AG, Allschwil, Switzerland) was equipped with a 125  1.6mm ODS-Hypersil, RP-18, 5-mm column (Bischoff, Leonberg, Germany). The mobile phase consisted of phase A [1 M TMAH, distilled water, and acetonitrile (20:880:100, v/v) adjusted to pH 4.5 with o-phosphoric acid] and phase B [1 M TMAH, distilled water, and acetonitrile (20:380:600) adjusted to pH 4.5 with o-phosphoric acid]. The program was a linear gradient from 27% phase B to 45% phase B in 9 min, with a ¯ow rate of 1.7 mL/min. The ultraviolet (UV) detection wavelength was 210 nm. Duplicate samples were assayed. Determination of SDZ EAA 494 and Its Degradation Products A 20-mg/mL solution of SDZ EAA 494 adjusted to pH 6.5 with sodium hydroxide was used for kinetic measurements. The concentrations of SDZ EAA 494 and its degradation products after thermal stress were determined by HPLC and a capillary electrophoresis (CE) methods. HPLC Assay. The HPLC system (consisting of Spectra Physics system Compaq 4/66, pump P4000, autosampler AS3500, and Focus UVDetector; Spectra Physics AG, Allschwil, Switzerland) was equipped with a 200  4.6-mm HyperJOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

280

MANN, KIEFER, AND LEUENBERGER

carb of Hypersil, porous graphitic carbon, 5-mm column (Runcon, Cheshire, England) as the stationary phase. The mobile phase consisted of phase A (5 mM KH2PO4 solution in distilled water, adjusted to pH 4.5 with o-phosphoric acid) and phase B [25 mM KH2PO4 and acetonitrile (900:100) adjusted to pH 4.0 with o-phosphoric acid]. The program was a linear gradient from 85% phase A to 0% phase A in 30 min, and then back to 85% phase A in 1 min, and holding this phase composition for 10 min. The ¯ow rate was 0.5 mL/min. The UV detection was set at 210 nm. Duplicate samples were assayed. CE Assay. The formation of the ( ‡ )-antipode during heat treatment was determined by CE with UV detection after derivatization. For derivatization, an aliquot of 1 mL of the samples was diluted a 20 mL with 0.1 M borate buffer (pH 9.5) and 0.75 mL derivatization reagent (8 mg/mL dansyl chloride dissolved in acetonitrile). The solutions were kept at 30 C for 30 min in the dark. After adding 100 mL of the stop reagent (0.04% aqueous methylenchloride solution) to each sample and shaking, these solutions were directly injected for CE analysis. Run buffer was made up as follows: 0.1 M borate buffer in distilled water and containing 3 mM g-cyclodextrin, adjusted to pH 9.5 with 2 M sodium hydroxide and ®ltered through a 0.45-mm pore ®lter. The CE system (HP-3D-CE, Hewlett Packard AG, Urdorf, Switzerland) was equipped with a bare fusedsilica capillary (total length, 60 cm; i.d., 75 mm). Injection was with an applied pressure of 50 mbar for 6 s. Voltage was applied at 250 V/cm. The UV detector was set at 214 nm. The percentage of the ( ‡ )-antipode was calculated using normalized areas. Duplicate samples were analyzed. Preparation and Growth of Biological Indicator Organisms Spores of B. stearothermophilus, derived from ATCC 7953, were cultured on TSA. Incubation was at 53.8  0.2 C for 72 h. The cells were harvested, and spores were grown on sporulation medium at 53.8  0.2 C for 4 days; a sporulation rate of 90% was determined by microscopic control. The sporulation medium contained 8.0 g of Nutrient Broth, 4.0 g of yeast extract, 20.0 g of Bacto Agar, and 968.0 g of distilled water and was supplemented with sterile MnCl2 solution to a concentration of 0.002% MnCl2 after sterilization. The spores were harvested and cleaned by cenJOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

trifugation (15 min, 1.4  g, 5 C) using a ChristMinifuge No. 40488 (Heraeus, ZuÈrich, Switzerland) and washing with WFI. To reduce the number of vegetative cells, the suspension was submitted to heat shocking for 10 min at 80 C and then cleaned by multiple centrifugations and washings with WFI. The spore stock suspension (9  108 spores/mL) was stored in WFI at 4 C. All experiments were carried out with the same spore crop. Preparation of Inoculated Formulations For the spore heat resistance experiments, 0.066 M phosphate buffers of pH 5.0 and 7.0 and a 0.1% aqueous CaCl2 solution (pH 6.0) in WFI were prepared. The pH values were controlled before and after sterilization of the formulations (121 C, 20 min) using a pH meter (model SA 520, Orion Research Inc., Cambridge, MA). The formulations were inoculated only shortly before the heat inactivation experiment. Therefore, an aliquot of the spore stock solution and centrifuged at 1:4  g at 5 C for 15 min. The supernatant was removed, and the spore pellets were resuspended in the sterile formulations to yield a concentration of 1  106 to 1  107 spores/mL. For heat inactivation under continuous ¯ow conditions, the inoculated formulations were poured into sterile, 200-mL glass bottles and immediately applied to heat inactivation experiments. The heat-stressed samples were collected in sterile tubes and kept at 4 C until further use. Viable Spore Counts All experiments were carried out at least twice. A plate count method was used to count viable spores in the samples before and after the heat treatment. Serial decimal dilutions of each sample in TSB were prepared and, from these dilutions, an aliquot of 1.0 mL was incorporated in TSA and plated. Colony-forming units (CFU) were counted after incubating for 48 h at 55± 57 C. The DT values were calculated as the negative reciprocal of the slope of the least square regression line of the semilogarithmic survival curve. The DT values were determined with the values of the linear portion of the survival curve. Only survival curves with a correlation coef®cient, r2 , of  0:96 and with more than four values in the linear portion of the inactivation curve were used. The logarithms of the DT values were plotted versus their corresponding temperature, and the

HIGH-TEMPERATURE SHORT-TIME STERILIZATION OF HEAT-SENSITIVE PRODUCTS

281

reciprocal of the slope of the least square regression line was taken as the z value. The reaction rate constants (k) for the microbial inactivation were calculated using the determined ‡ DT values (k ˆ 2:303=DT ). The Arrhenius graph was constructed by creating a semilogarithmic plot of the reaction rate constants versus the reciprocal of the absolute temperature. The activation energy was calculated from the slope of the least square regression line.

RESULTS Determination of Spore Inactivation Kinetics in HTST Range Thermal resistance of indicator spores suspended in WFI, 0.1% CaCl2 solution, and in phosphate buffers (pH 5.0 and pH 7.0) was investigated. Spore suspensions were heated at selected temperatures in the typical autoclave sterilization and HTST sterilization ranges (121±150 C; and in CaCl2 solution, 121±155 C, respectively) and the decimal reduction times for each temperature (DT values) as well as the reaction rate constants (k) were derived as described. The two models for describing the temperature coef®cients (Bigelow and Arrhenius models) were compared with respect to the prediction and extrapolation from typical sterilization to HTST sterilization temperature ranges. Therefore, the experimentally determined data (DT value and k) in the lower temperature range (121±130 C) were extrapolated to the HTST range, and the 95% con®dence intervals for the two different models were calculated. The DT and k values calculated by the two different models were than compared with the experimentally determined data. Figure 2 shows the prediction of the reaction rate constants (k), the DT values for the HTST temperature range, the corresponding con®dence limits for the respective models, as well as the experimental data. Figures 3 and 4 show the log DT values in different formulations plotted against their corresponding temperature. The data shown in the ®gures represent the mean of at least duplicate assays. In the pH 5 phosphate buffer, the spores were less resistant than in all other formulations used for the investigation. Thermal resistance of the spores increased in CaCl2 solution in comparison with WFI and phosphate buffers. This effect was consistent over the whole temperature range. In the standard sterilization temperature range

Figure 2. Spore heat resistance in WFI: comparison of the Arrhenius model and the Bigelow model. Thermal inactivation data extrapolated using the Arrhenius model (square, open) and 95% confidence intervals (dotted line), thermal inactivation data extrapolated using the Bigelow model (triangles, open)and the 95% confidence interval (lines) of the prediction, and the experimentally determined inactivation data (points, black) are shown.

(121±130 C), the spores showed higher heat resistance in pH 7 phosphate buffer than in pH 5 phosphate buffer. This effect became less pronounced in the higher temperature range. In the HTST temperature range, thermal resistance in pH 7 phosphate buffer was only slightly increased compared with thermal resistance in pH 5 phosphate buffer. Table 1 shows the temperature coef®cients (z values and Ea ) and correlation coef®cients (r2 ) determined by least square regression. The temperature coef®cients

Figure 3. Spore heat resistance in WFI, CaCl2 solution, and pH 5 phosphate buffer in the temperature range 121±155 C. Mean DT values of at least duplicate determinations and linear regression line are shown. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

282

MANN, KIEFER, AND LEUENBERGER

Figure 4. Spore heat resistance in pH 5.0 and 7.0 phosphate buffer in the temperature range (121± 150 C). Mean DT values of at least duplicate determinations and linear regression line are shown.

were calculated in the typical sterilization range (121±130 C), in the HTST range (140±155 C), and for the whole temperature range (121± 155 C). The calculated z values ranged from 7.58 in pH 7.0 phosphate buffer to 8.42 in pH 5.0 phosphate buffer. Activation energies ranged from 378.99 kJ/mol in pH 5.0 phosphate buffer to 420.99 kJ/mol in pH 7.0 phosphate buffer. The z values determined in typical sterilization range (121±130 C) were larger than those determined in the HTST sterilization range (140±155 C). The z values tended to decrease with increasing temperature in all formulations investigated. Chemical Degradation of Octreotide and SDZ EAA 494 in HTST Range The temperature values used for degradation studies were 80, 100, 120, 140, and 160 C, to cover the whole range for typical autoclave and HTST sterilization processes. Residence times Table 1. Parameter z value z value z value Ea (kJ/mol) a b

were chosen to yield suf®cient drug degradation for determination of degradation kinetics. The concentrations of Octreotide before and after heat treatment were determined by HPLC. During heat treatment, Octreotide was hydrolyzed to the major degradation product [Des-Throl8]-Octreotide. No degradation of SDZ EAA 494 could be detected by HPLC (data not shown). The same samples were analyzed by CE to determine the amount of ( ‡ )-antipode. The formation of the ( ‡ )-antipode was the only degradation product of SDZ EAA 494 that could be detected under the conditions investigated in this study. Both reactions, the hydrolysis and the racemization, could be described as ®rst-order reactions because the plots of ln (Ct /C0 ) versus time were linear. The ®rst-order reaction rate constants were calculated from the original concentration (C0 ) and the amount remaining (Ct ) at the time t [ln …Ct /C0 † ˆ ÿk  t]. The speci®c rate constants and correlation coef®cients (r2 ) of the least square regression line of both drug degradations were calculated at all temperatures investigated (Table 2). The logarithms of the rate constants were plotted against the reciprocal of their corresponding temperatures. Using the Arrhenius equation, the activation energy (Ea ) was found to be 83.93 kJ/ mol (r2 ˆ 0:98) for the hydrolysis of Octreotide and 113.5 kJ/mol (r2 ˆ 0:86) for the racemization of the SDZ EAA 494. The reaction rate and thus the degradation at any time±temperature process within the investigated temperature range could be calculated. The acceptable degradation limit during the sterilization process was assumed to be 0.5%. The process time, causing a 0.5% degradation of the drug substance, was calculated for the temperature range 80±160 C. The thermal resistance of B. stearothermophilus spores in the Octreotide and SDZ EAA 494 drug formulations was determined at 121, 122.5, and 140 C. The z value determined for the Octreotide formulation

Thermal Resistance of B. Stearothermophilus Spores in Different Formulations Temperature Range,  C 121±130 140±150 121±150

WFI 8.22 (r2 6.11 (r2 8.00 (r2 398.88 (r2

ˆ 0:98) ˆ 0:98) ˆ 0:99) ˆ 0:99)

CaCl2 Solution 8.17 7.72 8.32 383.55

z value over the temperature range 140±155 C. z value over the temperature range 121±155 C.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

(r2 (r2 (r2 (r2

ˆ 0:99) ˆ 0:94)a ˆ 0:99)b ˆ 0:99)

Phosphate Buffer (pH 5) 8.69 7.86 8.42 378.99

(r2 (r2 (r2 (r2

ˆ 0:96) ˆ 0:99) ˆ 0:99) ˆ 0:99)

Phosphate Buffer (pH 7) 7.57 5.71 7.58 420.99

(r2 (r2 (r2 (r2

ˆ 1:00) ˆ 0:96) ˆ 0:98) ˆ 0:98)

HIGH-TEMPERATURE SHORT-TIME STERILIZATION OF HEAT-SENSITIVE PRODUCTS

283

Table 2. First-order Reaction Rate Constants, Calculated for the Loss of Octreotide and SDZ EAA 494 Temperature ( C)

Octreotide Rate Constant (s ÿ 1)

r2

SDZ EAA 494 Rate Constant (s ÿ 1)

r2

80 100 120 140 160 Activation energy

0:012  10ÿ2 0:033  10ÿ2 0:180  10ÿ2 0:877  10ÿ2 1:473  10ÿ2 83.93 kJ/mol

0.77 0.62 0.61 0.91 0.96 0.96

0:0042  10ÿ2 0:0025  10ÿ2 0:0682  10ÿ2 1:3881  10ÿ2 1:3858  10ÿ2 113.50 kJ/mol

0.94 0.96 0.88 0.96 0.98 0.86

was 7.85 C, and the Ea was 406.50 kJ/mol (r2 ˆ 0:98). The z value in the SDZ EAA 494formulation was 9.26 C, and the Ea was 344.98 kJ/mol (r2 ˆ 0:99). From these data, the required exposure times causing a 12-log reduction of indicator spores in the product formulations were calculated and plotted on a semilogarithmic scale against their corresponding process temperature (Figure 5). In the same ®gure the process parameters (temperature time combinations) causing a 0.5% degradation of the heat sensitive drugs are shown.

Figure 5. Process parameters causing a 12-log reduction of B. stearothermophilus spores (dotted line, open squares) in Octreotide drug formulation and in SDZ EAA 494 drug formulation (dotted line, open triangles) and process parameters causing a 0.5% degradation (lines) of SDZ EAA 494 (black triangle) and Octreotide (black squares) are shown. Sterilization with acceptable drug degradation ( < 0.5%) can be performed at temperatures higher than the temperature where the line of the product degradation and corresponding microbiological effect intercept.

DISCUSSION Determination of Spore Inactivation Kinetics in the HTST Range For the design of an HTST sterilization process, more detailed knowledge about microbiological effects in the HTST range is needed. In the HTST sterilization process, the microbial effect during the heating and cooling phases contributes considerably to the total lethal effect of the process and residence time is in the range of seconds and milliseconds (12-log reduction of B. stearothermophilus at 150 C is 0.3 s). Therefore, for designing the HTST sterilization process, attention has to be paid to the temperature coef®cient used in this temperature range. The inactivation kinetics based on the Arrhenius concept and the Bigelow or z value model are quite different. Whereas one is a straight line, the other is curved. If the two models are compared, a frequent argument is made that the Arrhenius model is superior to the Bigelow model because the z value is temperature dependent per de®nition, whereas the activation energy (Ea ) is not. By describing the temperature dependence of the spore inactivation reaction according to the Arrhenius concept and the z value concept, differences in the times necessary for the inactivation of the spores occur. Those deviations in inactivation times calculated using these two different concepts depend on the reference temperature and the residence temperature selected for calculation. For example, the difference in residence times increases as the difference in reference temperature and residence temperature increases. Residence times calculated according to the Arrhenius concept are generally longer than residence times determined using the z value concept. In practice, the temperature dependence of the z value often is JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

284

MANN, KIEFER, AND LEUENBERGER

neglected. The z value is determined using the inactivation time versus temperature plot and considered as constant over the temperature range investigated.17 This simpli®cation is only acceptable over a narrow temperature range. If we examine the experimentally determined thermal resistance data, we ®nd that the z value decreases with increasing temperature in all formulations investigated (see Figures 2, 3, and 4, and Table 1). The z values determined in the HTST range were smaller than those in standard sterilization temperatures. In other words, the z value curves tend to be slightly concave downward. This result is in accordance with other studies carried out on the spores of the genus Bacillus, speci®cally of B. stearothermophilus and subtilis, showing a decreasing z value (temperature range investigated 120±140 C) as temperature increases.18 This phenomenon seems to start at temperatures in the range of 110 to 120 C. At temperatures < 110 C, z values of 10 C or larger are not uncommon, whereas at temperatures > 120 C, z values as small as 6 C have been reported.17 For microorganisms of the genus Clostridium, z value data tend to form straight lines (on a semilogarithmic plot) over a temperature span of 20 to 30 C. When the experimental data produce curves where the z value tends to be slightly concave downward, this situation is in opposition to the Arrhenius model which predicts a z value that increases with temperature. Figure 2 clearly shows that the experimental data better ®t the thermal inactivation kinetics extrapolated with the Bigelow model than those extrapolated with the Arrhenius concept. The Arrhenius model 95% con®dence interval did not comprise the experimental data. The Bigelow model seems to be more accurate in prediction of DT values in the HTST range. However, it has to be taken into consideration that extrapolation of microbiological inactivation data can only be done with true risk. The resulting curve will have wide con®dence limits because of the variability in microbiological results. When the z value and the Ea values can be determined directly from experimental data and not by extrapolation, the difference in residence times is generally much less pronounced. Considering the great uncertainty in measuring microbial inactivation data, there will be more variability in the individual data, whether it is called D or k, then the difference between the two models used to calculate the temperature coef®cient. On the basis of accuracy, the Bigelow and JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

Arrhenius models are essentially equal if experimentally determined data are used instead of extrapolation. For calculating the F value and designing the HTST sterilization process, attention has to be paid to the temperature coef®cient used in this temperature range. In a standard sterilization process, a z value of 10 C is used for the F value calculation.2 If the true z value in the HTST range is larger than the z value used, the lethal effect against microorganisms is smaller than assumed by the F value calculation. If the effective z value decreases with increased temperature, the F value calculation results in added security when the standard z value of 10 C is used. The z values determined over the whole temperature range varied between 7.58 and 8.42 C (Table 1) and tended to decrease with increasing temperature. The z values determined are within the expected span for B. stearothermophilus, which supports the view that the effective z value for B. stearothermophilus spores is in the range of 7.0 to 8.0 C.16 These results mean that using a z value of 10 C for process design, the lethal effect against microorganisms will be larger than indicated by the calculated F value in the process, especially in The HTST range where the z value tends to decrease. Added safety in the HTST process results if the standard z value is used instead of the effective z value. There is no hazard for the HTST process when the F concept with a z value of 10 C is used for the process design. For HTST process design, it is important to know whether the well-known in¯uences on indicator spore resistance can be extrapolated to the HTST range. It is well documented that spores are more easily inactivated at pH < 7.0 under typical sterilization conditions. The sensitivity of B. stearothermophilus and other spores to high concentration of hydrogen ions is well documented.19 A decrease in pH is often connected to a change of the z value. The pH 5.0 phosphate buffer effectively enhanced the thermal lethality on B. stearothermophilus spores compared with the pH 7.0 phosphate buffer. The comparison of the DT values of B. stearothermophilus in pH 5.0 and 7.0 phosphate buffers shows a great difference in typical sterilization temperatures, whereas the difference decreases at higher temperatures. In the HTST range, nearly no difference in the DT value is detectable. The z value in pH 7.0 phosphate buffer was smaller than that in pH 5.0 phosphate buffer. These data suggest that the in¯uence of pH on heat resistance of B.

HIGH-TEMPERATURE SHORT-TIME STERILIZATION OF HEAT-SENSITIVE PRODUCTS

stearothermophilus spores decreased in the HTST range. However, the decreased in¯uence of pH on heat resistance is not a phenomenon of HTST sterilization temperatures. The different z values in phosphate buffer with varying pH are apparent in the standard sterilization temperature. According to this difference in z value, the difference in the DT value between pH 5.0 and pH 7.0 phosphate buffers becomes smaller when the DT values are extrapolated to the HTST range. Thus, the difference in the DT value is more pronounced in the standard temperature range than in the HTST range because of the different z values in solutions of different pH. The decreasing in¯uence of pH is the result of extrapolating different z values to the HTST range. Bivalent cations are known to enhance the thermal resistance of bacterial spores. These cations are believed to chelate with dipicolinic acid of the spore coat to yield a higher heat resistance.20 Our data showed that spore heat resistance was increased in Ca2 ‡ solution compared with WFI consistently over the whole temperature range. Reports of decreased in¯uence of Ca2 ‡ ions and other chemical agents on spore resistance in the HTST range could not be con®rmed.21 Thus, the known in¯uence of changed pH and Ca2 ‡ ions on heat resistance of B. stearothermophilus spores is consistent over the whole temperature range. Chemical Degradation of Octreotide and SDZ EAA 494 in the HTST Range Knowing the degradation kinetics of the drug substances in the HTST range and the microbial inactivation kinetics in the same temperature range, the process parameters resulting in 0.5% drug degradation and the process parameters resulting in a 12-log reduction of indicator spores could be determined. Figure 5 shows the corresponding chemical and microbiological effects for both products, Octreotide and SDZ EAA 494. The dotted lines show the margin for the minimal thermal effect (time temperature combination) resulting in a 12-log reduction of indicator spores. All temperature time combinations above these lines result in a lethal effect higher than the 12log reduction and all temperature±time combinations below these dotted lines result in a smaller lethal effect than a 12-log reduction. The lines showing the time±temperature combinations resulting in 0.5% drug degradation are the margins for acceptable drug degradation during

285

a sterilization process. All process parameters (temperature time combinations) above these lines result in > 0.5% drug degradation and all process parameters below these lines result in < 0.5% drug degradation. The line showing the microbial effect in Octreotide formulation and the line of Octreotide drug degradation intercept at a process temperature of 145 C. If we focus on the microbiological effect, sterilization processes at < 145 C result in considerable drug degradation if a 12-log reduction is achieved. All sterilization processes with holding temperatures > 145 C achieve 12-log reduction but result in < 0.5% drug degradation. A sterilization process with 12-log reduction of indicator spores and acceptable drug degradation for the SDZ EAA 494 product is feasible above 158 C. The lines of the microbial effect and the drug degradation intercept at this temperature. For both heat-sensitive products, process parameters could be determined where sterilization is feasible with acceptable drug degradation ( < 0.5%). For a 6-log reduction cycle, the minimal temperatures for sterilization process with acceptable degradation rate are reduced to 140 C for Octreotide and to 153 C for SDZ EAA 494. The results demonstrate that the higher the process temperature, the less drug degradation occurs, whereas the lethal effect against indicator spores remains the same (12-log reduction). A higher process temperature would be expected to yield higher degradation levels. The fact that the sterilizing effect is achieved in a shorter time means that the product is exposed to a combination of temperature and time that causes less drug degradation. Similarly, although a low temperature cycle might be expected to yield lower levels of degradation, sterilization at these temperatures needs much longer to achieve the same microbial lethality. The resultant combination of temperature and time leads to higher levels of degradation. Degradation can be substantially reduced by employing high-temperature sterilization, provided the heating and cooling phases, where chemical degradation occurs without substantial lethal microbiological effect, are kept as short as possible. The Arrhenius equation can be used to estimate the destruction of both the drug substances in solution and microorganisms for any sterilization process. If the destruction rate of the drug substance, determined by the activation energy of the degradation process, is much smaller than the destruction rate associated with the bacterial JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

286

MANN, KIEFER, AND LEUENBERGER

inactivation, the reaction rate of bacterial inactivation will be much more in¯uenced by increased process temperature than the degradation of the drug substance. Thus, in the HTST process, adequate microbial inactivation can be obtained while unacceptable drug degradation may be avoided. The smaller the difference between the activation energies associated with the drug degradation and the activation energy associated with the microbial inactivation, the higher the process temperature required for a sterilization process with acceptable drug degradation. The difference in activation energy is much higher for the Octreotide than for the SDZ EAA 494 product formulation. This difference results in an adequate sterilization process with dwell temperatures higher than 145 C for Octreotide and higher than 158 C for SDZ EAA 494. The time±temperature combination, where a sterilization process results in acceptable product loss with suf®cient microbial effect, has to be determined separately for each product. Knowing the activation energy of the drug degradation and the activation energy or the z value of the microbial inactivation, process parameters resulting in acceptable loss of product can be calculated. If the drug degradation mechanism is consistent over the whole temperature range, parameters determined at standard sterilization temperatures can be extrapolated to the HTST range. The results of this study are also applicable for products undergoing zero-order or second-order degradation kinetics.1 HTST sterilization can only provide sterilization of the bulk solution. To prevent contamination during the subsequent ®lling of the sterile solution, other sterile manufacturing processes such as form-®ll-seal or isolator technology must be coupled with the HTST sterilization to assure the required sterility assurance level in the ®nal product.

CONCLUSIONS The two heat-sensitive products investigated, that can not be sterilized using typical autoclave sterilization, could be sterilized with acceptable drug degradation using HTST sterilization. The greater the difference in activation energies associated with the degradation of the drug substance and of the microbial inactivation, the lower was the minimal temperature needed for the HTST sterilization process with acceptable drug degraJOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001

dation. The comparison of the Arrhenius model and the Bigelow concept showed that the Bigelow model is more accurate to predict the thermal inactivation of B. stearothermophilus in the HTST range. The temperature coef®cient of the microbial inactivation reaction tended to decrease with increasing residence temperature in all formulations. The in¯uence of Ca2 ‡ ions and decreased pH on spore lethality is consistent over the entire temperature range according to the z value determined in the lower temperature range. The continuous-¯ow sterilizer, developed speci®cally for this study, ful®lled the requirements for kinetics measurements in the HTST range. The short heating and cooling time and the ¯exible exposure time enables one to determine both the microbial inactivation kinetics and the chemical degradation kinetics.

ACKNOWLEDGMENTS The authors thank Martin Strebel and Thomas Mannschott for their helpful advice and Dr. Dieter Witthauer for reviewing the manuscript.

REFERENCES 1. Martin A, Swarbrick J, Cammaratha A. 1987. In Stricker H, editor. Physikalische Pharmazie, 3rd edition. Stuttgart: Wissenschaftliche Verlagsgeselschft, pp. 197±228. 2. Marino F. 1993. Industrial Sterilization: A review of current principles and practices. In Avis KE, Lieberman HA, Lachman L, editors. Pharmaceutical Dosage Forms Parenteral Medications, Volume 2, 2nd edition. New York: Marcel Dekker, pp. 473±540. 3. Cook AP, MacLeod TM, Appleton JD, Fell AF. 1998. HPLC studies on the degradation pro®les of glucose 5% solutions subjected to heat sterilization in a microprocessor-controlled autoclave. J Clin Pharm Ther 14:189±195. 4. Mannermaa JP, Muttonen E, Liruusi JY, MaÈaÈttaÈnen L. 1992. The use of different time/temperature combinations in the optimization of sterilization of Ringers/glucose infusion solution. J Parent Drug Assoc 46:184±191. 5. Li LC, Parasrampuria J, Bommireddi A, Pec E, Dudleston A, Mayoral J. 1998. Moist-heat sterilization and the chemical stability of heat-labile parenteral solutions. Drug Dev Ind Pharm 24:89±93. 6. European Pharmacopeia. 1997. 7. Arrhenius S. 1887. Zeitung Physik Chem 1:631. 8. Bigelow WD. 1920. The logarithmic nature of thermal death time curves. J Infect Dis 29:528±536.

HIGH-TEMPERATURE SHORT-TIME STERILIZATION OF HEAT-SENSITIVE PRODUCTS

9. Stern AJ, Proctor BE. 1954. A micro-method and apparatus for the multiple determination of rates of destruction of bacteria and bacterial spores subjected to heat. Food Technol 8:139±143. 10. Geissler S. 1995. Optimierte WaÈrmeuÈbertragung bei kleinen Reynoldszahlen. Verfahrenstechnik 29:22±25. 11. Alfa Lava: Sterimedia Mini, Continuous Sterilization in Laboratory, Product Brochure. 1986. Alfa Laval, Lund, Sweden. 12. Charm E, Landau S, Williams B, Horowitz B, Prince AM, Pascual D. 1992. High-temperature short-time heat inactivation of HIV and other viruses in human blood plasma. Vox Sanguinis 62:12±20. 13. Gaze JE, Brown KL. 1988. The heat resistance of spores over the temperature range 120 to 140 C. Int J Food Sci Technol 23:373±378. 14. David JRD, Merson RL. 1990. Kinetic parameters for inactivation of bacillus stearothermophilus at high temperatures. J Food Sci 55:488±493. 15. Weast RC, Astle MJ, Beyer WH. 1986. Handbook of chemistry and physics, 66th edition. Boca Raton: CRC Press.

287

16. Ramayya SV, Antal MJ, Jr. 1989. Evaluation of systematic error incurred in the plug ¯ow idealization of tubular reactor data. Energy Fuels 3:105± 108. 17. P¯ug IJ. 1990. Microbiology and engineering of sterilization processes, 7th edition. Minneapolis: Environmental Sterilization Laboratory. 18. P¯ug IJ. 1996. Discussion of the z-value to use in calculating the F0-value for high temperature sterilization processes. PDA J Pharm Sci Technol 50:51±54. 19. Anderson RA, Friesen WT. 1974. The thermal resistance of Bacillus stearothermophilus spores, the effect of temperature and pH of the heating medium. Pharm Acta Helv 49:295±298. 20. Mallidis CG, Schole®eld J. 1987. Relation of the heat resistance of bacterial spores to chemical composition and structure I. Relation to core components. J Appl Bacteriol 62:65±69. 21. Santos MHS, Zarzo JT. 1997. The effect of ethylendiamintetracetic acid on the heat resistance and recovery of Clostridium sporogenes PA 3679 spores treated in HTST conditions. J Food Microbiol 34: 293±305.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 90, NO. 3, MARCH 2001