Determination of Thermal Inactivation Kinetics of Hepatitis A Virus in Blue Mussel (Mytilus edulis) Homogenate

Abstract

Hepatitis A virus (HAV) is a food-borne enteric virus responsible for outbreaks of hepatitis associated with shellfish consumption. The objectives of this study were to determine the thermal inactivation behavior of HAV in blue mussels, to compare the first-order and Weibull models to describe the data, to calculate Arrhenius activation energy for each model, and to evaluate model efficiency by using selected statistical criteria. The times required to reduce the population by 1 log cycle (D-values) calculated from the first-order model (50 to 72°C) ranged from 1.07 to 54.17 min for HAV. Using the Weibull model, the times required to destroy 1 log unit (t_{D = 1}) of HAV at the same temperatures were 1.57 to 37.91 min. At 72°C, the treatment times required to achieve a 6-log reduction were 7.49 min for the first-order model and 8.47 min for the Weibull model. The z-values (changes in temperature required for a 90% change in the log D-values) calculated for HAV were 15.88 ± 3.97°C (R², 0.94) with the Weibull model and 12.97 ± 0.59°C (R², 0.93) with the first-order model. The calculated activation energies for the first-order model and the Weibull model were 165 and 153 kJ/mol, respectively. The results revealed that the Weibull model was more appropriate for representing the thermal inactivation behavior of HAV in blue mussels. Correct understanding of the thermal inactivation behavior of HAV could allow precise determination of the thermal process conditions to prevent food-borne viral outbreaks associated with the consumption of contaminated mussels.

INTRODUCTION

Bivalve shellfish can become contaminated with viruses and other agents because they obtain their food by filtering small particles. In the process of filter feeding, bivalve shellfish may also concentrate and retain human pathogens derived from the environment (1). Epidemiological evidence suggests that human enteric viruses are the most common pathogens transmitted by shellfish. Hepatitis A virus (HAV) causes a severe viral infection linked to shellfish consumption, resulting in a serious debilitating disease and, occasionally, death (2). HAV can remain infectious within shellfish tissues for as long as 3 weeks (3). Even though the linkage of HAV infection to shellfish consumption was established approximately 50 years ago, HAV outbreaks associated with seafood have been, and currently remain, a serious public health concern. While an efficacious vaccine has reduced the overall incidence of HAV in the United States and elsewhere, shellfish-associated outbreaks still occur (1, 4–6).

HAV is a nonenveloped RNA virus, structurally similar to noroviruses, enteroviruses, and astroviruses. Numerous studies have addressed the high stability of HAV under denaturing environmental conditions relative to that of other nonenveloped RNA viruses (7). Due to its resistance to thermal treatment, a cell culture-adapted HAV strain would seem to be a relevant indicator in studies aimed at developing thermal inactivation strategies for most enteric viruses (8–11), especially since efforts to cultivate human norovirus have been unsuccessful (8, 9). However, it is very likely that more than one viral surrogate would be necessary in order to determine behaviors in response to different stressors or inactivation treatments. A few strains of HAV (HM-175, HAS-15, MBB 11/5) have been used for inactivation studies, are adaptable to cell culture, and can be maintained using fetal rhesus monkey kidney cells (FRhK-4) and/or human fetal lung fibroblasts (MRC-5) (12, 13).

Heating appears to be the most effective measure for the inactivation of HAV (7). In general, mussels or other shellfish are prepared by cooking, but generally they are heated only until the shells open, which is usually achieved at temperatures under 70°C for 47 ± 5 s (14). Clearly, shell opening is not indicative of whether the product has reached the recommended internal temperature. To ensure food safety, a minimum temperature and heating time are required, and these are independent of whether the shellfish has opened (15). Thus, shell opening is not a sufficient indicator for viral inactivation and does not ensure shellfish safety (4).

There have been studies on the thermal inactivation of HAV in seafood such as cockles (16), mussels (4, 17), greenshell mussels (15), and clams (18, 19). Although research has been done on the thermal inactivation of HAV in mussels, the reported results are inconsistent, most likely because of the thermal processing conditions utilized. For example, Millard et al. (16) reported that an internal temperature of 85 to 90°C for 1 min was sufficient to inactivate HAV in cockle meat (ca. 4 log 50% tissue culture infective doses [TCID₅₀]/ml). However, Croci et al. (4) concluded that immersion at 100°C for 1 min was not sufficient to inactivate HAV in homogenized Mediterranean mussels (Mytilus galloprovincialis) and that it was necessary to extend the heat treatment to 100°C for 2 min for complete inactivation (5.5 log TCID₅₀/ml). In another study, Hewitt and Greening (15) stated that thermal treatments at 90 to 92°C for 3 min were sufficient to achieve a 3.5-log reduction in the amount of HAV in New Zealand greenshell mussels (Perna canaliculus). Similarly, Hewitt and Greening (15) and Sow et al. (18) concluded that the application of 90°C for 3 min was sufficient to obtain a 5.47-log reduction in the amount of HAV in soft-shell clams (Mya arenaria). In contrast, Cappellozza et al. (19) reported that 90°C for 10 min was required to inactivate 5.43 log units of HAV in Manila clams (Ruditapes philippinarum). Due to the variable results in the literature, there appears to be a need to utilize precise thermal inactivation conditions to establish the minimal thermal processing conditions required to obtain a safe product.

Mathematical models to predict the thermal inactivation of food-borne pathogens assist in developing adequate thermal processes. Recent studies conducted on the thermal inactivation of human norovirus surrogates (20–23) revealed that the Weibull model was statistically superior to the first-order model in describing the thermal inactivation kinetics of norovirus surrogates. In order to provide a valid prediction, the determination of appropriate selection criteria and correct interpretation of those selected criteria are as important as model construction. The selection criteria used to determine goodness of fit include the coefficient of determination, correlation factor, predicted versus observed data, root mean square error (RMSE), and percentage of variance (24, 25). To our knowledge, there are no studies on the thermal inactivation kinetics of HAV in blue mussels (Mytilus edulis). The generation of precise thermal process data and the establishment of proper thermal processes for inactivating HAV in mussels would seem to be important both for consumers and for industry. Therefore, the purposes of this study were (i) to characterize the thermal inactivation behavior of HAV in blue mussels, (ii) to compare the first-order and Weibull models in describing the data in terms of selected statistical parameters, and (iii) to calculate z-values (changes in temperature required for a 90% change in the log D-values) and activation energy for each model.

MATERIALS AND METHODS

Viruses and cell lines.

Hepatitis A virus (HAV; strain HM175) and fetal monkey kidney cells (FRhK4) were kindly provided by Kalmia Kniel (University of Delaware). FRhK4 cells were maintained in Dulbecco's modified Eagle's medium–Ham's F-12 medium (DMEM-F12; HyClone Laboratories, Logan, UT) supplemented with 10% heat-inactivated fetal bovine serum (FBS; HyClone Laboratories, Logan, UT) and 1× Anti-Anti (antibiotic-antimycotic; Invitrogen, Grand Island, NY) at 37°C under an atmosphere containing 5% CO₂.

Propagation of viruses.

FRhK4 cells at ∼90% confluence in cell culture flasks were washed with phosphate-buffered saline (PBS; pH 7.4) twice before the addition of HAV stocks to its host cell monolayers. The infected cells were then incubated until >90% cell lysis in a water-jacketed CO₂ incubator at 37°C. Viruses were recovered by centrifugation at 5,000 × g for 10 min followed by filtration through 0.2-μm filters, aliquoted, and stored at −80°C until use.

Inoculation of mussels.

Fresh blue mussels (Mytilus edulis) were purchased from a local seafood market. The blue mussels were harvested from the North Atlantic coast during the winter season. The fresh mussel samples were shucked and homogenized using a Waring blender (model 1063; Waring Commercial, USA) at maximum speed. Since the primary objective of this study was to investigate the interaction of the virus and heat, homogenized blue mussel samples were used to obtain a uniform food matrix and a homogenous temperature distribution. Five milliliters of a virus stock (HAV) with an initial titer of 7.04 ± 1.34 log PFU/ml was added to 25 g of mussels in a sterile beaker, and the mixture was held at 4°C for 24 h. The inoculated blue mussel sample without heat treatment was used as a control, and viruses were enumerated.

Thermal treatment.

Heat treatment was carried out in a Haake model V26 circulating water bath (Thermo Haake, Karlsruhe, Germany) at selected temperatures ([50 to 72°C] ± 0.1°C) for different durations (0 to 6 min) in 2-ml vial glass tubes. Sterilized vials (2 ml) were carefully filled with inoculated homogenized mussels by using sterile pipettes in a biosafety cabinet. The filled vials were rinsed in 70% ethanol before immersion in a thermostatically controlled water bath. The water bath temperature was confirmed with a mercury-in-glass (MIG) thermometer (Fisher Scientific, Pittsburgh, PA) and by the placement of type T thermocouples (Omega Engineering, Inc., Stamford, CT) in the geometric center of the water bath. Another thermocouple probe was placed at the geometric center of a vial through the vial lid and in contact with the mussel sample to monitor the internal temperature. The thermocouples were connected to an MMS3000-T6V4 type portable data recorder (Commtest Instruments, Christchurch, New Zealand) to monitor the temperature. Samples were heated at 50, 56, 60, 65, and 72°C for different treatment times (0 to 6 min). The times required to reach the target temperature (come-up times [CUT]) at 50, 56, 60, 65, and 72°C were 104, 113, 154, 166, and 187 s, respectively. The treatment time began (and was recorded) when the target internal temperature reached the designated temperature. A sample was taken to enumerate HAV prior to heating, when the sample reached the target temperature (t = 0), and at all sampling time points. Triplicate tubes were used for each time and temperature point. After the thermal treatment, sample vials were immediately cooled in an ice water bath for 15 min to stop further thermal inactivation. The vial contents were collected in a sterile beaker by using a sterile pipette. The remaining contents of the vials were washed with elution buffer by using sterile pipettes to flush out the entire sample, and the virus extraction protocol was followed.

Inoculated mussels without heat treatment were used as controls and HAV enumerated.

Virus extraction.

The virus was extracted as described by Baert et al. (14) with some modifications. Virus was extracted from the inoculated sample and enumerated before thermal treatment and following each thermal treatment. Inoculated and thermally treated mussels were washed with 12.5 ml of elution buffer (ratio, 1:6) containing 0.05 M glycine (BP381-5; Fisher Scientific, USA) and 0.15 M NaCl (S671-500; Fisher Scientific) at pH 9.0 to allow the detachment of virus particles from the food matrix in the presence of an alkaline environment. The pH was then adjusted to 9.0 using 10 M NaOH (S80-45; Sigma-Aldrich,USA). Samples in the sterile beaker were put on a shaking platform (120 rpm) and were kept for 20 min at 4°C. Samples were centrifuged at 10,000 × g for 15 min at 4°C (in an Eppendorf centrifuge, model 5804R, USA), and the pH of the supernatant was adjusted to 7.2 to 7.4 using 6 N HCl (Sigma-Aldrich, H17-58) to improve the PEG precipitation of the virus particles. Polyethylene glycol (PEG) 6000 (Fisher Scientific, A17541-0B) and NaCl were added to obtain a final concentration of 6% PEG (wt/vol) and 0.3 M NaCl. These samples were first placed on a shaking platform (120 rpm) overnight at 4°C and then centrifuged at 10,000 × g for 30 min at 4°C (in a model 5804R centrifuge; Eppendorf, USA). The supernatant was discarded, and the pellet was dissolved in 2 ml PBS and was put on a shaker for 20 min to homogenize. Virus extracts were stored at −80°C until the enumeration of plaques using HAV plaque assays.

Enumeration of survivors by infectious plaque assays.

Thermally treated and control viral suspensions were diluted 1:10 in DMEM-F12 containing fetal bovine serum (2%) and 1% antibiotic-antimycotic. The infectivity of each treated virus in comparison to that of untreated virus controls was evaluated by standardized plaque assays following the procedures described previously by Su et al. (26). Viral survivors were enumerated as PFU per milliliter.

Modeling of inactivation kinetics. (i) First-order kinetics.

The first-order kinetic model assumes a linear logarithmic reduction of the number of survivors over treatment time. The model for the inactivation of microorganisms can be written as follows:

$$-\frac{dN}{dt}=kN$$ (1)

where N is the number of survivors at an exposure time (t) and k is the first-order rate constant (per minute). The integration from time zero to time t yields equation 2:

$$\frac{N_{(t)}}{N_0}=e^{-kt}$$ (2)

where N₀ is the initial number of microorganisms. The slope of the survival curve will always be a straight line with slope k. The time to reduce the population by 1 log cycle (D-value [D]) is expressed by equation 3:

$$D=\frac{\ln \hspace{0.17em}(10)}{k}=\frac{2.303}{k}$$ (3)

Combining equation 3 with equation 2 yields the first-order survival model,

$${\log \hspace{0.17em}}_{10}\frac{N_{(t)}}{N_0}=-\frac{t}{D}$$ (4)

where N_(t) is the number of survivors after an exposure time (t) and N₀ is the initial population (both expressed in PFU per milliliter). D is the decimal reduction time in minutes (time required to kill 90% of viruses), and t is the treatment time (in minutes).

(ii) Weibull model.

The Weibull model assumes that the survival curve is a cumulative distribution of lethal effects

$$\frac{N_{(t)}}{N_0}=\exp \hspace{0.17em}\left[-{\left(\frac{t}{\propto }\right)}^{\text{β}}\right]$$ (5)

where α (per minute) and β are scale and shape parameters for the Weibull model, respectively.

As indicated by previous studies (27), if one takes the inverse of the scale factor (α) as a reaction rate constant k′ (per minute), the equation becomes

$$\frac{N_{(t)}}{N_0}=\exp \hspace{0.17em}\left[-{\left(k\prime t\right)}^{\text{β}}\right]$$ (6)

For the Weibull model, the time required to reduce the number of microorganisms by 90% (analogous to the D-value) can be calculated by using shape and scale parameters (28) as shown in equation 7:

$$t_D=\propto {\left[-\ln \left({10}^{-D}\right)\right]}^{1/\beta }$$ (7)

where D represents 90% reduction of a microbial population. t_D is valid only when it refers to the treatment time starting at time zero.

Arrhenius activation energy.

The inactivation rate is influenced primarily by temperature, and the temperature dependence of the rate constant is typically described by the Arrhenius equation:

$$k=A\exp \hspace{0.17em}\left(-\frac{E_a}{RT}\right)$$ (8)

where A is a frequency factor which is constant, E_a is the activation energy (in joules per mole), R is the universal gas constant (8.314 J mol⁻¹ K⁻¹), k is the rate constant (1/min), and T is the absolute temperature (K).

The inactivation rate constants obtained for each model were then fitted to an Arrhenius equation.

$$\ln \hspace{0.17em}k=\ln \hspace{0.17em}A-\frac{E_a}{RT}$$ (9)

By plotting lnk versus 1/T, the slope of the curve will be a straight line which equals the activation energy. This concept has been used to calculate the activation energy of microbial inactivation (29).

Statistical analysis.

Statistical and nonlinear regression analyses were performed using the SPSS statistical package, version 11.0.1. The statistical criteria applied to distinguish among the kinetic models were R² (coefficient of determination), r (correlation coefficient), root mean square error (RMSE) (the lower the better), and standard errors (SE) for each coefficient. In addition to r, RMSE, and SE, the percentage of variance (%V) is accounted for by the model (based on the number of terms):

$$\%V=\left[1-\frac{\left(1-R^2\right)\left(n-1\right)}{n-N_T-1}\right]\times 100$$ (10)

where R² is the coefficient of determination, n is the number of data points, and N_T is the number of model equation terms. This coefficient takes into account the complexity of the model and the population of data used to describe it. As the number of observations n increases, the number of terms (N_T) has less of an effect on the fitness of the model.

The confidence level used to determine statistical significance was 95%.

RESULTS AND DISCUSSION

After the inoculation of blue mussel samples with HAV stocks, the titer of unheated virus recovered was 6.73 ± 1.27 log PFU/ml of homogenate for the control. The survival curves of HAV in blue mussels at different temperatures (50 to 72°C) are shown in Fig. 1. As the temperature and/or treatment time increased, virus inactivation also increased. During the CUT, lethality also occurred. At 50°C, the amount of reduction was 0.43 log PFU/ml during CUT. There was an increase in the log reduction during CUT with increasing temperature, to a maximum of 1.1 log PFU/ml during CUT at 72°C. As stated by Chung et al. (30), the size of the heating vessel contributed to differences in the heat transfer rate and resulted in different CUT times for the same microorganism strain. Since the reduction in the number of survivors during CUT is important for the determination of precise thermal process conditions, the CUT should be taken into account in designing appropriate thermal processes. The shapes of the inactivation curves were characterized by an initial drop in viral counts followed by a tailing behavior. Visual inspection of these survival curves indicated that a nonlinear model would describe these data better than a linear model (Fig. 1).

FIG 1.

Survival curves of hepatitis A virus in blue mussels treated at 50°C (A), 56°C (B), 60°C (C), 65°C (D), and 72°C (E). Open circles, first-order model; black circles, experimental values; gray circles, Weibull model.

The D-values calculated from the first-order model (50 to 72°C) were in the range of 54.17 ± 4.94 to 1.07 ± 0.24 min (Table 1). The temperature had a significant effect on D-values for the temperature range studied (P, <0.05). To understand the relationship between inactivation rate and temperature, it is necessary to examine the underlying inactivation mechanism during thermal treatment. Hirneisen et al. (31) stated that the mechanism of heat inactivation of viruses involves changes in the capsids of virus particles. The virus capsid is the protein coat that encloses the viral genome and any other components necessary for virus structure or function and is also responsible for binding to the host. Croci et al. (32) stated that exposure to mild temperatures (ca. 50°C) leads mainly to damage to the viral receptor binding site through structural changes in the capsid protein, which then does not allow binding and thus causes low levels of inactivation. Higher inactivation rates at increased temperatures (>56°C) may be due to denaturation of capsid proteins. At higher temperatures, alteration of tertiary structure occurs, and therefore, the capsid does not play a protective role against the degradation of nucleic material (33). This hypothesis is supported by previous research (4, 18, 22, 23, 32, 34, 35).

TABLE 1.

Coefficients of the first-order and Weibull models for the survival curves of HAV in mussels during thermal inactivation

T (°C)	Weibull distribution				First-order kinetics
	β	α (min−1)	tD = 1 (min)	R2	D (min)	R2
50	1.02 ± 0.45	16.91 ± 9.38	37.91 ± 6.95	0.91	54.17 ± 4.94	0.89
56	0.50 ± 0.02	1.97 ± 0.04	10.43 ± 0.49	0.99	9.32 ± 3.26	0.91
60	0.65 ± 0.09	2.13 ± 0.33	7.73 ± 0.20	0.99	3.25 ± 0.72	0.90
65	0.43 ± 0.02	0.96 ± 0.13	6.73 ± 0.30	0.99	2.16 ± 0.17	0.86
72	0.53 ± 0.19	0.32 ± 0.05	1.57 ± 1.04	0.96	1.07 ± 0.24	0.91

Although several studies have been performed to investigate the thermal inactivation of HAV in mussels (4, 15, 17), no studies on the calculation of thermal inactivation parameters were found. While valuable empirical information was gathered, no thermal kinetic information was generated, and thus, it is impossible to design an adequate thermal process within the limits of those studies.

Temperature had a significant effect on both t_D and D-values for the range from 50 to 72°C (P, <0.05). The parameters of the Weibull model (the scale factor [α] and shape factor [β]) were used to calculate the t_D value, which was used as an analog to the D-value of the first-order model (Table 1). For the Weibull model, the time required to destroy 1 log (t_{D = 1}) of HAV ranged from 37.91 ± 6.95 to 1.57 ± 1.04 min for the temperature range from 50 to 72°C. This indicates overprocessing (if the target is 1 log unit) if the first-order model is used instead of the Weibull model. However, a 1-log reduction as a target is rarely seen in food processing; therefore, it is wise to compare 6-log reduction timess (as in the case of pasteurization). Calculation of the time needed for a 6-log reduction (often used as a target for processes such as pasteurization) by the first-order model yields 6 times the D-value (6D). However, the time needed for a 6-log reduction by the Weibull model is not 6t_{D = 1} but t_{D = 6}; this is the consequence of nonlinear behavior. Based on the thermal inactivation data obtained from the present study, at 72°C, the required treatment time to achieve a 6-log reduction of HAV was 7.49 min by the first-order model and 8.47 min by the Weibull model. This indicates underprocessing if the first-order model is used instead of the Weibull model. These under- and overprocessing phenomena could explain the impressive safety record of the first-order model for many years, especially in the canning industry, where overprocessing for Clostridium botulinum is widely practiced (29).

For further investigation of the Weibull model, the temperature dependence of the parameters may be evaluated. The range of the Weibull shape factor (β) for the temperatures studied (50 to 72°C) was 1.02 ± 0.45 to 0.53 ± 0.19; the range of the Weibull scale factor (α) was 16.91 ± 9.38 to 0.32 ± 0.05 min⁻¹. van Boekel (28) reviewed 55 thermal inactivation studies on microbial vegetative cells and concluded that in most cases, shape factors were clearly independent of the heating temperature whereas scale factors could be a function of the heating environment. Thus, a change in the scale factor described the effect of the heating environment on inactivation. The present study is consistent with those findings in that temperature did not influence the shape parameter. In this study, at 50°C, monotonic downward concave (shoulder) behavior was observed with a shape factor of 1.02 ± 0.45. At temperatures higher than 50°C, monotonic upward concave (tailing) behavior was observed with shape factors ranging from 0.43 ± 0.02 to 0.65 ± 0.09, but no relationship was observed between the temperature and the shape parameter. However, the scale parameter was dependent on heating temperature. A second-order polynomial model was established to quantify the influence of temperature on the scale factor. The relationship between the scale factor and temperature is expressed by the following equation:

$$\mathbf{\propto }=0.0651{\mathit{T}}^2\hspace{0.17em}-8.5887\mathit{T}\hspace{0.17em}-282.26$$

where T is the temperature (°C) and R² is 0.902.

The temperature dependencies of the inactivation rate constants (k and k′) were fitted to the exponential Arrhenius function for both models (Table 2). The Weibull model gave a higher R² than the first-order model when the rate constants were fitted to the Arrhenius equation (Table 2). For the first-order model, the estimated inactivation rate constants ranged from 0.04 ± 0.01 to 2.21 ± 0.44 min⁻¹ for the temperature range of 50 to 72°C. The temperature dependency of the inactivation rate constant by the first-order model is expressed by a second-order polynomial model:

$$\mathit{k}=0.0037{\mathit{T}}^2-0.3579\mathit{T}\hspace{0.17em}+8.6101$$

where R² is 0.99.

TABLE 2.

Arrhenius inactivation rate constants of the first-order and Weibull models for the survival curves of HAV in mussels during thermal inactivation

T (°C)	First-order model		Weibull model
	k (min−1)	R2	k (min−1)	R2
50	0.04 ± 0.01	0.88	0.08 ± 0.05	0.92
56	0.27 ± 0.11	0.87	0.51 ± 0.01	0.99
60	0.63 ± 0.13	0.83	0.48 ± 0.07	0.99
65	1.07 ± 0.09	0.79	0.65 ± 0.05	0.99
72	2.21 ± 0.44	0.86	3.18 ± 0.50	0.97

The estimated inactivation rate constants by the Weibull model for the temperatures studied (50 to 72°C) were 0.08 ± 0.05 to 3.18 ± 0.50 min⁻¹. A second-order polynomial model was established to quantify the influence of temperature on the inactivation rate constant for the Weibull model. The relationship between the inactivation rate constant and temperature was as follows:

$$\mathit{k}\prime \hspace{0.17em}=0.0079{\mathit{T}}^2-0.8385\mathit{T}\hspace{0.17em}-22.306$$

where R² is 0.97.

When living organisms at the microscopic level are exposed to heat, they do not all receive the same dose of energy per unit of time. For an inactivation event to occur, the interacting molecules need a minimum amount of energy, the activation energy (36). This energy causes denaturation in the target organism. According to the first-order model, there is a log linear relationship between the energy required for inactivation and the temperature. The calculated activation energies for the first-order model and the Weibull model based on inactivation rate constants are shown in Fig. 2. The activation energy obtained from the first-order model was 165 kJ/mol, while for the Weibull model, the activation energy was 153 kJ/mol. Inactivation curves with shoulders and/or tailing are usually explained by the concept that more than one critical target may require more than one hit before being inactivated.

FIG 2.

Arrhenius plot of the HAV inactivation rate constant versus temperature for the first-order model (A) and the Weibull model (B). (R², 0.93 for both models).

In addition to activation energies for both models, z-values were also calculated. The z-values for HAV were 15.88 ± 3.97°C (R², 0.94) using the Weibull model and 12.97 ± 0.59°C (R², 0.93) for the first-order model (Fig. 3). Using the parameters generated in the present study, an industrial thermal process for whole mussels could be estimated. The reported CUT for whole New Zealand greenshell mussels (Perna canaliculus) in boiling water (100°C) was 240 s (15). According to Stumbo (37), the contribution of the CUT (t_c) to the apparent lethality of a process can be calculated by the addition of 0.4 · t_c (in minutes) to the calculated process time for the specific temperature used. For whole mussels, on the basis of the CUT determined by Hewitt and Greening (15) and the thermal inactivation parameters obtained from the present study, the process times required to achieve a 6-log reduction of the HAV population in mussels in boiling water (100°C) would be 2.7 and 3.2 min for the first-order and Weibull models, respectively. It should be noted that the use of alternative heat processes might result in slightly different heating characteristics, and thus the effectiveness of a thermal process must be validated prior to the actual application of that process.

FIG 3.

Thermal death time curves for hepatitis A virus in blue mussels by the Weibull model (R², 0.94) (A) and the first-order model (R², 0.93) (B).

To compare the goodness of fit of the first-order and Weibull models, the coefficient of determination (R²), correlation factor (r), RMSE, and percentage of variance (%V) were calculated (Table 3). The Weibull model consistently produced the best fit for all the survival curves. For the survival curves at 50 to 72°C, the Weibull model had R² values of 0.91 to 0.96, correlation factor values of 0.95 to 0.99, RMSE values of 0.01 to 0.04, and percentages of variance of 88 to 95%. Accurate model prediction of survival curves would be beneficial to the food industry in selecting the optimum combinations of temperature and time to obtain the desired levels of inactivation. The present results revealed that the Weibull model could be used successfully to describe thermal inactivation of HAV in blue mussels.

TABLE 3.

Statistical comparison of the first-order and Weibull models for the survival curves of HAV

Model	T (°C)	RMSE	R2	r	%V
First order	50	0.05	0.89	0.94	85
	56	0.14	0.91	0.95	88
	60	0.11	0.90	0.95	87
	65	0.14	0.86	0.93	81
	72	0.18	0.91	0.95	88
Weibull	50	0.04	0.91	0.95	88
	56	0.01	0.99	0.99	99
	60	0.01	0.99	0.99	99
	65	0.01	0.99	0.95	99
	72	0.04	0.96	0.98	95

Conclusion.

The heat resistance of HAV was greatly affected by temperatures from 50 to 72°C. The application of higher temperatures likely caused the denaturation of both nucleic material and capsid protein, resulting in significantly decreased D and t_D values. The z-values obtained from the first order and Weibull models were 12.97 ± 0.59°C (R², 0.93) and 15.88 ± 3.97°C (R², 0.94), respectively. The calculated activation energies for the first-order model and the Weibull model were 165 and 153 kJ/mol, respectively. Precise information on the thermal inactivation of HAV in mussels was generated, enabling the calculation of parameters for more-reliable thermal processes to control and/or inactivate the virus in potentially contaminated mussels and thus prevent food-borne illness outbreaks.