Klijn NIZO Food Research, PO Box 20, 6710 BA Ede, The Netherlands.
Aims: We discuss several factors that are critical for heat inactivation experiments and which should be taken into account for future research.
Methods and Results: On the basis of examples from the literature we discuss critical factors influencing the calculated heat inactivation of Mycobacterium avium subsp. paratuberculosis (MAP). Furthermore, using a modelling approach, we show that tailing of the inactivation curve of MAP is caused by the presence of cell clumps and not by a more heat-resistant cell fraction.
Conclusions: The experimental conditions of the MAP heat inactivation studies of different research groups vary significantly and lead to considerable differences in results and conclusions. Therefore, a more consensual approach should be employed in future studies. In addition, our model on clumping of MAP can be used to predict the decimal reduction of MAP during heat treatment and to study the effect of clumping on other lethal effects.
Significance and Impact of the Study: We discuss several factors that should be carefully considered in heat resistance experiments. This is essential for a thorough interpretation of results from experiments and should be given proper attention in future experiments and publications on this topic.
Heat treatment is the process most often applied to guarantee the quality and safety of milk and milk products. The heat treatment or pasteurization of milk derives its principles from the work of Louis Pasteur (1822–1985) who developed a method to prevent abnormal fermentation in wine by destroying the responsible organisms by heating to 60°C. Since 1880, cow’s milk given to babies has been treated to reduce the risk of infection by heating in a continuous flow to temperatures of about 60–75°C (Staal 1986). In most developed countries pasteurization of milk for consumption became compulsory in the first half of the 20th century (Staal 1986), in particular to control the spread of Mycobacterium bovis from cattle to humans. Mycobacterium tuberculosis was considered to be the most heat-resistant human non-spore-forming pathogen associated with milk. In 1957, the recommended holding time for low-temperature holding pasteurization was increased to ensure effective killing of Coxiella burnetii cells.
Recently, however, concern has been expressed about the efficacy of milk pasteurization in killing M. avium subsp. paratuberculosis (MAP). Millar et al. (1996) found MAP-positive polymerase chain reaction signals in pasteurized retail milk and suggested the possible survival of this bacterium during pasteurization conditions currently applied in practice. Mycobacterium avium subsp. paratuberculosis is the aetiologic agent of Johne’s disease, a chronic intestinal infection in cattle and other ruminants. The clinicopathological features of Johne’s disease are very similar to those of Crohn’s disease, a severe chronic inflammation of the gastrointestinal tract of humans. Since MAP has been isolated from the intestinal tract of human patients with Crohn’s disease, it is suggested that MAP might be associated with this disease (Chiodini et al. 1984).
Because of the possible link between MAP and Crohn’s disease, several research groups have studied the survival of MAP under laboratory pasteurization conditions and published data on the heat inactivation of MAP. The interpretation and comparison of these data are quite complicated due to the different experimental conditions employed by the investigators. Since many researchers still perform experiments to collect heat inactivation data for MAP (i.e. D- and z-values), it is important to consider all critical parameters that influence heat inactivation experiments, as they are of utmost importance for the interpretation of results of MAP pasteurization experiments. The effect of heat treatment on the inactivation of bacteria, spores and enzymes has been described in several publications (Freeman et al. 1968; Wilkinson and Davies 1973; Ordonez et al. 1987; MacDonald and Sutherland 1993; Peck et al. 1995; Breand et al. 1997; Huemer et al. 1998; Laurent et al. 1999). Previously, specific models were developed in order to predict the heat inactivation rate of micro-organisms based on their intrinsic heat resistance value (Ea, activation energy) and temperature dependency value (z-value, the temperature difference required for a 10-fold decimal reduction time) (Hermier et al. 1975). However, these values can be affected by several factors that are critical in heat inactivation experiments.
In this article we will discuss critical factors influencing the calculated heat inactivation of MAP. In view of this, the data generated to date and in future can be used to develop an improved model to predict the heat inactivation kinetics of MAP.
CRITICAL PARAMETERS IN HEAT INACTIVATION EXPERIMENTS
Strains and culture conditions
An important factor in determining heat inactivation of bacteria is the type of strain used in the experiments, as heat resistance is strain dependent. In addition, other factors such as continuous subculturing of strains in the laboratory and the growth phase of the cultures may have even more impact on the heat resistance of bacteria. Subculturing of Bacillus sporothermodurans, for instance, results in formation of less heat-resistant spores as compared with those from freshly isolated field strains (Huemer et al. 1998). According to Jay (1992), bacterial cells are less heat resistant while in the logarithmic phase than in the stationary phase of growth. For MAP it is suggested that the difference in heat resistance (D-values) among strains in heat inactivation experiments are primarily dependent upon the level of subculturing and not on the strain (Sung and Collins 1998). Low-passage MAP clinical strains seem more sensitive to killing by heat treatment than high-passage laboratory strains (Sung and Collins 1998). For M. avium subsp. avium, which is closely related to MAP, it has been shown that organisms grown in vitro were more heat resistant than organisms of the same strain grown in vivo (Merkal et al. 1981).
Not only the growth conditions prior to heat treatment, but also those after heating will influence survival of bacterial cells. After heat treatment many cells will be sublethally injured, as can be observed by the substantial delay in detecting growth of heat-treated bacterial cells compared with unheated cells (D’Aoust 1978; Meylan et al. 1996; Keswani and Frank 1998; Semanchek et al. 1999). Hence, recovery of injured MAP cells in an appropriate medium after heat treatment is very important and may influence MAP survival counts (Damato and Collins 1990; Keswani and Frank 1998). Liquid media seem to be more advantageous to the recovery of sublethally injured cells than solid media (Collins et al. 1988; Sockett et al. 1992). The resuscitation of MAP cells after heat treatment in a liquid medium followed by plating on a solid medium may, therefore, result in higher colony counts.
Application of heat
The heat inactivation of bacteria is very dependent on the way in which the heat is applied. The temperature range and time scale (the total amount of heat/energy provided) are also critical parameters (Cerf and Griffiths 2000). The legislation for foods only covers the definitions of pasteurization and sterilization. The International Dairy Federation defines pasteurization as ‘a process by heat treatment with the aim of minimizing possible health hazards arising from pathogenic micro-organisms and extending keeping quality by reducing the number of spoilage micro-organisms consistent with minimal chemical, physical and organoleptic changes in the product’ (Asperger 1993). In fact, precise values for temperature and time are not given.
In dairy practice most heat treatments are performed in a continuous and turbulent flow process at high temperatures for short times (HTST). In such a process milk is preheated, followed by the actual heat treatment and subsequently rapidly cooled to 20 or 4°C. In contrast, however, in most laboratories MAP heat inactivation experiments were performed as a batch process at low temperatures for long holding times, usually followed by a quick cooling on melting ice (Chiodini and Hermon-Taylor 1993; Grant et al. 1996a; Grant et al. 1996b; Stabel et al. 1997; Keswani and Frank 1998). For example, experiments with B. sporothermodurans have shown that batch heating in tubes for less than 1 min results in less inactivation of the cells than occurred under the same conditions in a UHT (Ultra High Temperature) pilot equipment (Huemer et al. 1998). This clearly illustrates that the way heat is applied may influence the inactivation of cells.
To obtain reliable data to calculate the heat inactivation kinetics of micro-organisms, multiple data points need to be examined. Analysis of only one end-point, as employed in most published experiments, will not provide sufficient data to calculate correctly the survival and inactivation kinetics (D- and z-value) of micro-organisms during heat treatment (Chiodini and Hermon-Taylor 1993). At each temperature at least three measurements are needed to calculate the decimal reduction time (Dt) adequately. Furthermore, the decimal reduction time needs to be determined for at least three different temperatures, as performed for MAP by Sung and Collins (1998). Only then can a reliable z-value be calculated and the inactivation kinetics determined for all time–temperature combinations.
Quantification of colony-forming units
One of the most important factors in estimating the survival of micro-organisms in milk after heat treatment is a reliable determination of colony-forming units (cfu) present in the milk. Unfortunately, the experimental end-point of culturability in conventional laboratory media is weak for determining the true presence or absence of residual viable MAP in a sample. This is not only because of the ‘viable-but-unculturable-state’, but also because of the general difficulty of accurately reporting MAP in cow’s milk by culture even when it has not been heat shocked at all. Moreover, MAP is notorious for its clumping during growth. Clumping also occurs in nature, as can be seen under the microscope in faecal samples. This may result in an underestimation of the total cell count by a factor of 100–1000. Heat treatment of MAP clumps followed by a rapid cooling may even result in an increased number of colonies after pasteurization (Chiodini and Hermon-Taylor 1993). This is explained by the breaking up of the MAP clumps by the high heat shock and subsequent rapid cooling on ice.
EVALUATION OF MYCOBACTERIUM AVIUM SUBSP. PARATUBERCULOSIS HEAT INACTIVATION DATA OBTAINED FROM THE LITERATURE
Table 1. Mycobacterium avium subsp. paratuberculosis. (MAP) inactivation experiments performed by different research groups
INACTIVATION MODEL FOR MYCOBACTERIUM AVIUM SUBSP. PARATUBERCULOSIS HEAT TREATMENT
Clumping has a great impact on the calculated heat inactivation of MAP. Clumping of cells can be incorporated in an inactivation model where the number of counted cells is described as a function of the absolute number of cells and the number of clumps. A clump will be counted as one cell (1 cfu) in a culturing experiment on solid media, no matter whether it consists of one or more living cells. The initial average number of bacteria per clump (Cclump,0) depends on the initial absolute number of cells (N0), the fraction of bacteria in clumps (f) and the number of clumps (nclumps)
It is assumed that, during heat inactivation, the inactivation of cells is equally spread over free cells and cells in clumps. Until the average number of cells per clump is less than one, the counted number of cells after heating time t will be the sum of the free cells and number of clumps
where Nt is the absolute number of cells after t seconds and Nfree,t is the number of free cells after t seconds. Nfree,t can be described as a function of heating time (t) and temperature (T)
The way in which the inactivation constant k(T) is affected by temperature is important in determining the extent of the overall inactivation after heat treatment. Although in the dairy industry several relations are used, particularly for the destruction of micro-organisms, the most appropriate description of the temperature dependence is given by the Arrhenius relationship
where ln k0 is the pre-exponential factor, Ea is the activation energy and R is the gas constant (8·314 J mol–1 K–1). When the average number of cells per clump becomes less than one, the number of counted cells equals the absolute number of cells
This model can be used to predict the (counted) decimal reduction in MAP during heat treatment and to study the effect of clumping on the determination of heat inactivation. In Fig. 1, the crucial role of clumping in heat inactivation is demonstrated; the number of surviving bacteria is plotted as a function of the heating time in the case that 0, 50, 70, 90 or 100% of the bacteria are in clumps, the initial number of MAP cells is 2 × 106 and the average number of cells per clump is 10·000 (it is clear that cell clumps in naturally contaminated milk will all have different sizes). Using the heat inactivation kinetics calculated from the experimental data of Sung and Collins (1998) (Ea=326·9 kJ mol–1, ln k0=113·2), the number of surviving bacteria can be determined (Figs 2 and 3). This shows that, with an increasing number of clumped bacteria, tailing will occur. This is in agreement with the observations of Grant et al. (1996a), Stabel et al. (1997) and Keswani and Frank (1998). Hence, the tailing effect has less to do with a more heat-resistant cell fraction but is most likely to be caused by a fraction of clumps with a high number of cells. In this fraction, even after an inactivation of four decimal reductions, living cells are still present in the clumps. This was also shown in micrographs of stained residual live cells in heat-treated clumps (Grant et al. 1996c; Grant et al. 1997). In Fig. 2, the counted decimal reduction is plotted as a function of the heating time and fraction of MAP in clumps, while the average number of MAP cells per clump is set at 10·000 and the heating temperature at 62·5°C. Moreover, it will be difficult to interpret the results of MAP heat inactivation experiments if only a few measurements are performed. In that case, insufficient data points are available to draw an accurate inactivation curve. Assuming that MAP is always (partly) present in clumps, it is clear that it is not possible to draw any conclusions with respect to heat inactivation and survival rate if only end-point measurements are performed as, based on the fraction of clumps, the calculated decimal reduction can vary significantly. For instance, the decimal reduction following heating for 10 min at 62·5°C can vary from 3 to 8.
It should be noted that this model does not account for the preheating and cooling step before and after the pasteurization step. During preheating a considerable amount of bacteria can be killed, resulting in a significantly altered D-value of MAP (Sung and Collins 1998). In contrast, rapid cooling after the heating step was shown to result in more surviving cells: the faster the cooling rate the larger the number of cells surviving (Chiodini and Hermon-Taylor 1993).
Several research groups have studied the survival of MAP during heat treatment of milk. The results obtained for three strains that have been used by at least two different research groups have been summarized in Table 1. It is surprising to see that, although the groups used different experimental conditions, the results are quite similar (considering natural spreading), i.e. D65°C=48–78 s, D63°C=1–1·5 min and D62°C=3·5–4·5 min. Unfortunately, until now only inactivation kinetics data (Dt values) have been presented, obtained by holder pasteurization in tubes and capillaries but not by flow pasteurization in (small-scale) industrial pasteurization units (Grant et al. 1996a; Keswani and Frank 1998; Sung and Collins 1998). Several publications describe experiments in which HTST equipment was used (Hope et al. 1996; Stabel et al. 1997) but, in these publications, only end-point measurements were performed. End-point measurements, however, cannot be used to calculate reliable inactivation kinetics. The inactivation in all heat processes performed in the dairy industry can only be calculated based on Ea or z-values. The time–temperature combinations for one, four or eight decimal reductions in MAP were calculated using the data of Sung and Collins (1998). In Fig. 3, the processing conditions are plotted for different milk products. As mentioned before, variation in pasteurization conditions can be found not only between different products but also between different countries. For pasteurized consumption milk this can vary from 15 s at 72°C in the UK to 30 s at 78°C in Italy. From Fig. 3 it can be concluded that, based on the currently available data, the inactivation of MAP during industrial pasteurization is at least four to more than eight decimal reductions.
As outlined in this article, the experimental conditions of the MAP heat inactivation studies of the different research groups vary significantly and lead to considerable differences in results and conclusions. The most striking difference is that some groups observe a tailing effect in the thermal inactivation curves of MAP whereas colleagues do not. We have explained that this is a result of bacterial cell clumps present in the milk matrix. This clumping phenomenon makes the enumeration of MAP quite complicated. Moreover, for a complete assessment of the survival of MAP in pasteurized milk products, the level of contamination in raw milk needs to be determined. It can be expected that the level of contamination can vary considerably in different regions or countries, based on the prevalence of clinical cases of Johne’s disease, and may also be subject to seasonal influences. It has been demonstrated that clinical cases of paratuberculosis make the largest contribution to the contamination rate of raw milk (Sweeney et al. 1992). Unfortunately, it is still very difficult to culture MAP quantitatively from milk and it is, therefore, almost impossible to obtain a reliable calculation of the numbers of MAP cells in raw and pasteurized milk. So far, only limited information is available on the level of MAP contamination of raw milk. In one publication the amount of culturable MAP cells in milk of non-clinically ill cows has been estimated at 40–80 cells l–1 (Sweeney et al. 1992). In view of this limited data, if it is assumed that pasteurization of milk results in a four to eight decimal reduction in MAP, the presence of viable MAP will vary from merely one cell per 1000 to one cell per 10 000 000 l packages of pasteurized milk. An evaluation of whether the four to eight decimal reductions that can be expected will be enough to produce sufficient inactivation of MAP in pasteurized milk and milk products can only be made when the actual natural MAP contamination is known.
At present, essential information is still lacking on the level of MAP in naturally contaminated raw milk, on the heat inactivation of MAP field strains and on reliable MAP inactivation data obtained from heat treatment experiments at temperatures between 70 and 80°C. This information is essential to make a good assessment of the inactivation by commercial pasteurization processes, in particular when they need to be changed to increase current inactivation rates.
Apart from heating time and heating temperature other factors in dairy processing, such as preheating (termization), centrifugation, bactofugation and homogenization, may also affect the survival of MAP during pasteurization. Throughout these processes, shearing forces may have a detrimental effect on bacterial cells and may make them less heat resistant. All these processes are used in the production of pasteurized retail milk in Europe. Unfortunately, until now no research has been performed on the effect of mechanical stress on MAP cells in relation to their heat resistance.
Many questions still need to be answered before a thorough risk assessment can be made of the safety aspects of MAP and its potential survival in pasteurized milk and milk products.
The authors thank Steef Biesterveld, Leo Langeveld and Jan Wouters for helpful discussions and for critically reviewing the manuscript.