An experiment to validate the precooking of tuna as a control for histamine formation was carried out at a commercial tuna factory in Fiji. Albacore tuna (Thunnus alalunga) were brought on board long-line catcher vessels alive, immediately chilled but never frozen, and delivered to an on-shore facility within 3 to 13 days. These fish were then allowed to spoil at 25 to 30°C for 21 to 25 h to induce high levels of histamine (>50 ppm), as a simulation of “worst-case” postharvest conditions, and subsequently frozen. These spoiled fish later were thawed normally and then precooked at a commercial tuna processing facility to a target maximum core temperature of 60°C. These tuna were then held at ambient temperatures of 19 to 37°C for up to 30 h, and samples were collected every 6 h for histamine analysis. After precooking, no further histamine formation was observed for 12 to 18 h, indicating that a conservative minimum core temperature of 60°C pauses subsequent histamine formation for 12 to 18 h. Using the maximum core temperature of 60°C provided a challenge study to validate a recommended minimum core temperature of 60°C, and 12 to 18 h was sufficient to convert precooked tuna into frozen loins or canned tuna. This industrial-scale process validation study provides support at a high confidence level for the preventive histamine control associated with precooking. This study was conducted with tuna deliberately allowed to spoil to induce high concentrations of histamine and histamine-forming capacity and to fail standard organoleptic evaluations, and the critical limits for precooking were validated. Thus, these limits can be used in a hazard analysis critical control point plan in which precooking is identified as a critical control point.
The principles of hazard analysis and critical control point (HACCP) plans (42) require the use of science-based evidence to validate a preventive process control as capable of effectively limiting an identified food hazard. The food safety regulation for the production of seafood products published by the U.S. Food and Drug Administration (FDA) (38) is based upon HACCP principles. This article describes an in-plant, full-production-scale validation process for ascertaining that precooking tuna to a minimum core temperature of 60°C pauses histamine (scombrotoxin) formation for at least 12 to 18 h, thus giving processors enough time to process fish of any size.
All seafood sold for human consumption in the United States must be processed according to the FDA's seafood HACCP regulation (21 CFR 123) (38). To facilitate compliance, the FDA has provided extensive guidance to the industry in the form of a document, “Fish and Fishery Products Hazards and Controls Guidance” (42). This FDA HACCP guidance provides the strategies, i.e., critical control points (CCPs) and critical limits (CLs), for control of various hazards associated with seafood products. Alternative approaches for controlling hazards at CCPs must be proven valid by the processor. The FDA seafood HACCP regulations mandate that processors “verify that the HACCP plan is adequate to control food safety hazards that are reasonably likely to occur” (38).
Fast-moving scombroid fishes, such as tuna, tend to have more free histidine in their muscles than do other fishes; therefore, higher concentrations of histamine and other amines can form (21, 32). Histamine will be formed in these fish when they are not chilled properly after capture. Histamine-forming bacteria (HFB) are ubiquitous in the marine environment, and under unchilled conditions, HFB in tuna can grow. The HFBs produce the enzyme histidine decarboxylase (HDC), which converts free histidine within the tuna muscle into histamine (3). Histamine is a relatively small, very stable toxin and is not inactivated by cold or heat (15).
Histamine and other amines are the primary cause of scombrotoxin reactions, often called histamine poisoning (18). Histamine poisoning is the leading cause of seafoodborne illness in the United States and internationally (7, 14). The defect action concentration for histamine in tuna flesh in the United States is 50 ppm (37), and the toxic concentration is 500 ppm (37).
No histamine is found in freshly caught tuna (16, 21), but when these fish are not handled properly at sea, histamine will form and accumulate. Chilling and/or freezing the fish promptly after capture will prevent the growth of HFB and histamine formation (42). The tuna fishing vessels in all major fisheries either chill the tuna with ice or refrigerated seawater (RSW) or freeze the tuna at sea with a blast freezer or brine freezing system (6, 17). Diligent time-temperature controls throughout the supply chain are essential to assure consumer safety because tuna fishing and processing are primarily done in the tropics, with typical ambient temperatures that could potentially lead to spoiled tuna with high histamine concentrations.
The basic method for tuna canning has been described by Bell et al. (4) and DeBeer et al. (8). Tuna loins are the desired tuna portions for canning. To facilitate skinning, deboning, and separating the red meat from the loin meat, the fish are initially heat treated in a process known as precooking (8). Tuna processors sometimes inject vegetable broth into the raw tuna loins to add flavor and improve texture. This vegetable broth is a flavoring allowed in the U.S. canned tuna standard of identity (40).
The FDA HACCP guidance (42) sets limits for the time allowed for processing tuna to 12 h from the time the tuna starts thawing until it has been sealed in cans and the retort has started or the loins start to freeze, if the ambient temperature exceeds 21°C at any time. Procedures that must be conducted in this time window include thawing, precooking, cooling, removing the skin and bones, separating the red meat from the loins, and preparing the fish for canning or freezing. This 12-h window is not sufficient for thawing, precooking, and cooling larger tuna. However, the HACCP guidance does allow an intermediate heat step when processors can demonstrate that histamine formation can be delayed long enough after precooking to allow additional processing time, i.e., to “restart the clock” (8, 42). Although most processors believe that precooking provides an adequate process control for safely producing canned tuna without elevated histamine concentrations, no scientific studies have been conducted using the “worst-case” conditions to validate the use of precooking to restart the clock (F.A., personal communication).
FDA warning letters (41, 43) and the subsequent collaboration between the FDA and the National Fisheries Institute identified a specific lack of data for demonstrating the effect of precooking on controlling the formation of histamine. One such warning letter (41), in October 2010, stated there was no precooking step listed in the existing HACCP plan as a CCP “to ensure control of exposures to operations to control scombrotoxin formation.” The warning letter went on to say, “the HACCP plan should include the pre-cook as a critical control point and should include all appropriate critical factors to achieve an adequate cook.” Therefore, precooking CCP and CL studies are clearly needed.
Enache et al. (13) conducted thermal death time studies on five species of known HFB (Morganella morganii, Raoultella planticola, Hafnia alvei, Enterobacter aerogenes, and Photobacterium damselae) and found that M. morganii was the most heat resistant; therefore, a heat treatment designed to control M. morganii would effectively control the other HFB. Nolte et al. (25) further developed a model demonstrating that once the tuna reaches a core temperature of 60°C, M. morganii populations have undergone a significant 5-log reduction. In other studies, the activity of HDC produced by M. morganii was reduced by 56% at 50°C and by 99% at 60°C (20). Osborne and Bremer (27) used the Bigelow model to determine the thermal death times for M. morganii in kahawai (Arripis trutta). That model indicated that the minimum precooked core temperatures should be 58 to 62°C to ensure that the final product would not produce histamine during subsequent temperature abuse (5, 27). The most common target end point internal product temperature (EPIPT) for precooked tuna for the past 50 years (F.N., personal communication) has been a minimum meat temperature at the backbone of 135°F (57.2°C), as developed by Peterson in 1971 (30).
The validation of a HACCP CL involves testing to determine whether the CL can control the worst case of the hazard. In this article, the worst case for tuna spoilage refers to tuna that has been subjected to deliberate time-temperature abuse to reach exponential histamine formation, thereby providing practical evidence of HFB presence in the test fish. The test fish also needed to fail standard organoleptic evaluations and exceed the FDA guidelines for histamine (50 ppm), thus indicating the presence of HFB.
The objectives of this study were to (i) precook worst-case spoiled tuna to a target maximum core temperature of 60°C to validate the precooking process as stopping or pausing histamine formation and (ii) determine and validate how much extra processing time is therefore gained from precooking the tuna to a minimum core temperature of 60°C by sampling tuna at different periods after precooking to determine whether histamine formation has been paused for a significant period.
MATERIALS AND METHODS
Design of the validation experiment
After comprehensive consultations with experts and statisticians from the FDA, a set of validation experiments was designed to accomplish the two study objectives. Tuna were to be brought on board the fishing vessels alive, spoiled to obtain histamine concentrations higher than 50 ppm, and then precooked to a targeted maximum of no higher than 60°C at the core to determine whether histamine formation could be paused for sufficient time to process larger tuna (at least 12 to 18 h after precooking). For purposes of validation, fish were deliberately spoiled before precooking to simulate time-temperature abuse on the catcher vessels and to satisfy the burden of proof required under the FDA's seafood HACCP regulation (38).
All the experimental albacore tuna (Thunnus alalunga) were caught by longline fishing vessels near Fiji and brought on board alive. Fish were immediately chilled on board in ice or RSW and stored at −2 to 0°C. The 226 albacore used in this experiment had a mean weight of 15 kg each.
All fresh raw tuna were landed and then spoiled in Suva, Fiji, on two separate days. The spoilage process started immediately after the tuna were unloaded, and each fish subsequently was spoiled to greater than 50 ppm of histamine. During the spoilage process, the fish were stored in open fish bins containing seawater for up to 25 h with temperatures that were controlled at 25 to 33°C. The method of deliberate spoilage of these fish is described in the Supplemental Material. After spoilage, the fish were completely frozen in air blast freezers (Dalian Refrigeration Company, Dalian, People's Republic of China) set at −40°C and held for at least 24 h. The spoiled frozen fish were then transferred by freezer truck (−20°C) to the processing facility.
Thawing and processing
All precooking experiments were conducted in precookers (Leonard Engineering, El Cajon, CA) at a commercial tuna processing facility in Levuka, Fiji. Two precooking experiments were conducted each day for 3 days. Standard methods for commercial tuna processing were used for thawing and evisceration to prepare the fish for precooking. The fish were thawed for 11 h to a target core temperature of −2 to 0°C in unchlorinated single-use seawater. The thawing water temperature was approximately 26 to 28°C. The thawed fish were butchered by removing the stomach, liver, heart, and other entrails. Some of the round fish were subsequently split. The heads and gills were not removed from the round or split tuna; however, the heads and gills were removed from the split injected fish. In this article, round fish means unsplit tuna fish with the stomach, liver, heart, and other entrails removed.
In all, 226 fish and 1,112 individual samples were used for histamine analysis to validate this precooking procedure. The design of experiment (DOE) stipulated six precooker treatment combinations (A through F) using 32 fish each (Table 1). Six fish pieces were tested after precooking at each of five sampling times for each treatment. Each fish piece was uniquely identified throughout the entire processing cycle. This DOE matrix of precooker treatments allowed for comparisons among the factors: whole round fish, split fish, split fish injected with vegetable broth (round fish were not injected for technical reasons), and cooking at the two precooker target steam temperatures (70 and 100°C).
Before precooking, the round fish (treatments A and B) were placed directly into fish baskets, with one fish per basket, and then into precooker racks. The round fish controls (treatment G), which were raw (without precooking treatment), were also placed into fish baskets and precooker racks in the same way so that these control fish could accompany the precooked fish throughout the rest of the process after precooking.
The fish for the noninjected treatments C and D were split in half using the butchering split saw and placed into fish baskets in single layers (two pieces per basket), and the baskets were placed into precooker racks. The fish for the split-injected treatments E and F were split in half, injected with vegetable broth using commercial injectors, and placed into fish baskets and fish racks in a manner similar to that for the split noninjected tuna. One half of each individual fish in the four split fish treatments (C, D, E, and F) was precooked, and the other half served as an uncooked raw control. The split fish halves were randomized such that the left or right halves were randomly assigned to either the precooked or raw control group. Therefore, the only difference between experimental groups and the controls was the precooking heat treatment, allowing for strong paired statistical analysis.
The fish from the two chilling methods aboard the longline catcher vessels (ice and RSW) were evenly distributed across the precooker treatments such that any potential differences could be identified during the statistical analysis. Of the six fish for each treatment sampling time, one had been chilled in ice and five had been chilled in RSW.
Precooker target temperatures and validation
Commercial precookers using direct contact steam were used to precook the fish. Precooker temperatures of 70 and 100°C were selected to bracket the range of steam temperatures used in commercial precooking schedules. This design allowed for comparison of potential differences resulting from precooking times and temperatures typically encountered in routine commercial procedures, e.g., lower precooker chamber temperatures requiring longer cooking times versus higher precooker chamber temperatures with shorter cooking times (F.N., personal communication). Temperature distribution studies were conducted to verify a uniform temperature distribution inside each precooker (F.N., personal communication) (24). For each of the precooker treatments, the precooker was filled with additional fish to replicate commercial precooker operations and temperature distribution patterns. Approximately 6 metric tons of additional nontest fish were processed per precooking treatment, for a total of 36 metric tons of nontest fish or about 2,400 individual nontest fish.
The core temperatures for all precooked test fish and a sample of the nonprecooked control fish were monitored with standard thermal probes (type T thermocouples, Omega Engineering, Norwalk, CT) connected to Calplex data loggers (TechniCAL, Metairie, LA) (33). The temperatures of the thawing water and sidespray were monitored every 30 min with programmable MiniVACQ wireless temperature data loggers (TMI-Orion, Castelnau-le-Lez, France) (34). Similarly, the ambient air temperatures in the areas for thawing, preparation, sidespray, conditioning, and packing were monitored with a MiniVACQ recording device every 30 min. The probes were verified for accuracy against a calibrated NIST traceable (National Institute of Standards and Technology, Gaithersburg, MD) thermometer at 0°C (ice slurry), 100°C (boiling water), and 70 and 100°C set points inside the precookers. The dial thermometers were calibrated against an NIST traceable thermometer. Temperature data were compiled and evaluated using Calsoft (TechniCAL), MiniVACQ (TMI-Orion), and Excel (Microsoft, Redmond, WA) software.
The frozen spoiled fish were thawed for 11 h, precooked for 2 to 5 h, and then cooled for 2 to 4 h in the sidespray and conditioning rooms to reach core temperatures below 40°C.
Raw fish controls
All raw fish controls, including the uncooked halves of the fish from treatments C through F and the round fish controls (G) for treatments A and B, were set aside in the preparation area where the ambient temperature was 26 to 32°C while the test fish were being precooked. All uncooked raw fish controls were treated in the same manner as the precooked fish throughout the remaining steps of the process in the sidespray cooling, conditioning, and packing rooms.
Precooking test fish
All test fish, except the raw controls, were precooked as outlined in the experimental design (Table 1) to a target precooker exit core temperature near but not exceeding 60°C. Temperature probes connected to the Calplex data loggers allowed close monitoring of fish core temperatures in real time during precooking. Three free leads were also placed near the precooker racks containing the precooked test fish to monitor the precooker steam temperature at the top, middle, and bottom positions of the rack.
As soon as the core temperature of any fish in the precooker treatment approached the target of 60°C, the steam was turned off, and all fish, including the nontest fish, were removed from the precooker as quickly and safely as possible. Fish core temperatures were measured using dial thermometers and recorded immediately for all 32 precooked test fish per treatment.
Sample collection time
For each treatment, six histamine samples were collected after the core temperatures were recorded (T0). The other precooked fish and uncooked raw controls were moved immediately into the sidespray room, and cooling with a water spray commenced to lower the precooked test fish core temperatures to the range where histamine concentrations would be expected to increase most rapidly (15). When the average core temperature reached 45°C, the fish were moved into the conditioning room for a brief time. After the average precooked fish core temperatures further cooled to 40°C in the conditioning room, the precooked and control fish were moved into the packing room to challenge the tuna through exposure to ambient conditions of 27 to 32°C.
Samples for baseline histamine analysis before precooking, designated B4PC, were collected from the raw fish controls and the treatment fish approximately 30 min before precooking. After precooking, histamine samples were collected from six fish per treatment, plus the corresponding baseline control fish at each of five sampling times: immediately (T0), 12 h (T12), 18 h (T18), 24 h (T24), and 30 h (T30).
Sample collection stratification
The precooking process resulted in the desired maximum 60°C exit core temperatures as planned except in a few cases. The steam portion of the precooking process was stopped before the core temperature of any fish in a treatment reached 60°C, but the fish core exit temperatures were not collected until the fish had been removed from the precooker (after 5 to 10 min). The core temperature continues to increase for a time after the steam has been turned off due to residual heat transfer or overshoot (8, 29). The 100°C precooker steam treatments tended to have greater increases in fish core postprecooking temperatures than did the 70°C precooker steam treatments because more residual heat remained within the outer tissues of the fish cooked at higher temperatures. This limitation of some fish having exit core temperatures higher than 60°C was addressed by selectively stratifying the fish across the five sampling times. T0 was selected for the fish with the highest core temperatures, because it was already known that the effect of fish core temperature on the histamine concentration and its increase is less important right after precooking than later. This selective stratification was done to ensure that all the lower exit temperature fish were sampled at the later sample times to determine the effect of lower exit temperatures on histamine concentrations in precooked fish over time. After stratification, the remaining fish were randomized using fish core temperature results and sampled across the five sampling times.
Samples for histamine analysis (minimum sample weight of 250 g) were collected from the following locations on each fish piece (Fig. 1): (i) for round fish, anterior dorsal and anterior ventral, right and left side, four samples per fish; and (ii) for split fish, anterior dorsal, anterior ventral, and middle ventral, three samples per split fish piece. Each sample location on each fish can be tested only once because histamine analysis is destructive. The anterior ventral location was chosen to comply with the recommendations from the FDA HACCP guidance regarding chemical testing for histamine (42), the anterior dorsal location was chosen because of historical industry practices based on research conducted by Frank et al. (16), and the middle ventral sample location was suggested by the FDA experts for an extra sample site on the split fish. Both the left and the right sides of all fish were sampled to further account for a possible nonuniform distribution of histamine throughout the fish (12, 22). To preserve the samples for laboratory analysis, all samples collected for histamine testing were immediately frozen using a commercial plate freezer (19) and analyzed using the AOAC fluorometric method 977.13 for testing histamine in seafood (1).
Sensory evaluations were conducted by sensory experts trained using the National Oceanic and Atmospheric Administration and FDA sensory training protocols (36) on each fish at the same time as samples were collected for histamine testing.
The DOE was a three-factor full factorial repeated-measures design with two treatments omitted (J.C., personal communication) (23, 26) and a covariate. The factors are (i) round and split, (ii) injection, and (iii) precooking steam temperature. Sampling time after precooking (0, 12, 18, 24, and 30 h) was the covariate. The DOE was for repeated measures (26) because each fish piece was measured at three or four locations depending on whether the fish was split or round (the three or four samples collected per fish piece were analyzed at the same time).
The DOE called for equal numbers of fish in each of the six precooking treatments. The histamine concentrations were converted to natural logarithms (ln) for the factorial analysis of variance (ANOVA) model statistical analysis. The ln form reduces the impact of outliers and helps satisfy the normality assumption for the ANOVA (J.C., personal communication).
The histamine results were analyzed with Minitab software (State College, PA) and Microsoft Excel. The mean ln histamine concentrations at each sampling time for each treatment group were analyzed using a one-way ANOVA. The differences in the histamine concentrations between the sampling times within a treatment were evaluated using Tukey's method for multiple comparisons of means (35).
The ln histamine results also were analyzed with a regression model to determine the comparative impacts of round versus split, injection, precooker temperature, and sampling time after precooking. The terms were eliminated from the model with the main effects and two-way interactions using a backwards elimination approach of terms that were not statistically significant. A quadratic term for sampling time was included to predict the time point when the histamine concentrations would exceed the initial concentrations.
The temperatures of the air, water, and fish cores during fish spoilage are shown in Supplemental Figure S1. The water temperature during spoilage was 25 to 32°C, and fish core temperatures increased from less than 4°C, when stored on ice or in RSW, to equilibration with these ambient water temperatures during the 21- to 25-h spoilage times (Fig. S1). Ten fish were tested for histamine as indicators to stop the spoilage time. The histamine concentrations were 66 to 214 ppm (Supplemental Table S1) before freezing the rest of the intact fish. The fish were also noticeably decomposed, as determined by sensory evaluations. Subsequently, every fish tested throughout this study had a histamine concentration higher than 50 ppm and had odors of decomposition. The histamine concentrations at the end of spoilage and after thawing but before precooking are shown in Figure 2. An added data point of 0 ppm of histamine based on historical data (16, 21) was included in Figure 2 for reference to represent the lack of histamine in the fresh fish at the start of spoilage.
Time-temperature exposure and precooking results
The average core temperature profile for the precooked test fish throughout the entire process is shown in Figure 3 and includes the stages in order of progression: start of thawing, end of thawing, precooker initial, precooker exit, cooling during sidespray, and conditioning room. The ambient temperatures in these processing areas were 25 to 35°C except in the conditioning room, which averaged 20°C (Fig. 4). All ambient room temperatures were in the optimal temperature zone for histamine formation (42).
The resulting EPIPTs are charted in an interval plot in Figure 5 and tallied in Table S2. The data in Table S2 are grouped by degree and precooker treatment. The target for the core precooker exit temperatures was achieved: most fish had core temperatures lower than 60°C, and only 11% exceeded the target core temperature maximum of 60°C.
Histamine concentrations and statistical analysis
The effects of fish core exit temperature, sampling time after precooking, and the three analysis factors on the mean ln histamine concentrations are shown in Table S3. Sampling time (hours after precooking stopped) was the most significant factor in predicting histamine concentrations in precooked fish. The effect of sampling time on ln histamine concentrations was nearly identical for the six treatments studied (interval plot, Fig. 6). The core fish exit temperatures were significantly associated with histamine concentrations (t = −4.97, P = 0.000). For every 1°C increase in the core fish exit temperature, there was a 2.1% decrease in ln histamine concentration at any time after precooking in these high-histamine fish (Table S3).
The differences in histamine concentrations over time for the six treatment combinations versus the controls are shown in Figure 7. The ANOVA results for raw versus precooked fish are shown in Table S4 and were significant (df = 1, F = 198.06, P = 0.000).
The mean histamine concentrations (and standard deviations) for all precooking treatment and sample time combinations are shown in Table 2. The large standard deviations indicate high variability in histamine concentrations for particular treatment and sample time combinations. An ANOVA was performed on each treatment group to test for differences in mean histamine concentrations across sampling times. Tukey's multiple comparison method (35) was used to identify which sampling times were significantly different.
The ANOVA results (Table 2) with Tukey's method indicate that the mean histamine concentrations in the precooked tuna at time points B4PC, T0, T12, and T18 were not significantly different from the concentrations at previous time points for any treatments. The mean histamine concentration at T24 was significantly different from concentrations at these earlier sampling times for treatments B, D, and E. The mean histamine concentration at T30 was significantly different from that at the earlier sampling times for all treatments. In summary, no significant differences in mean histamine concentrations were found until at least T24, even in these severely spoiled and decomposed fish.
Based on calculations using the coefficients derived from the regression model to fit a quadratic equation to the data, on average the histamine concentrations started to increase just after 18 h (18.7 h) after the fish were removed from the precooker. Figure 8 shows the quadratic function (P = 0.000).
High variability in histamine concentrations between fish was observed throughout the study. Significant differences in histamine concentrations were found between individual fish treated identically (i.e., different fish ID numbers; df = 150, F = 15.90, P = 0.000) (Table S5). Three or four sample locations were analyzed for histamine concentrations for each piece of precooked fish, but because the measurements within a single fish are correlated, these samples are not independent. The repeated measures aspect of this study was handled by analyzing the summary statistics from each fish, i.e., the mean ln histamine concentration. A second summary statistic, the maximum ln histamine concentration, was also analyzed; these results were essentially identical to those for the mean ln histamine concentration.
Significant differences in histamine concentrations were found between sample locations on the fish pieces (Table S5). The anterior dorsal location had the lowest mean histamine concentrations, the anterior ventral location had the next highest, and the middle ventral location had the highest (df = 2, F = 32.83, P = 0.000). This location effect was accounted for in the analysis by analyzing the locations separately, and the results were essentially identical to those obtained for the mean ln histamine concentrations. No significant differences were found between the left and the right sides of the fish (side factor in Table S5).
Effects of chilling method
The method of chilling, i.e., RSW or ice, was tracked throughout the study, but no significant differences in histamine concentrations were found for fish treated with these two onboard chilling methods (df = 1, F = 1.043, P = 0.311).
Effects of precooking temperatures
Two interactions were significant: between round fish and precooker temperature (df = 1, F = 7.14, P = 0.008) and between injected fish and precooker temperature (df = 1, F = 5.90, P = 0.016) (Table S6). The precooker temperature for the round × split interaction indicates that for round fish, the higher precooker temperatures resulted in lower histamine concentrations, whereas for the split fish, precooker temperature did not have a significant impact on histamine concentrations. The temperature × injection interaction indicates that for split injected fish, the higher precooker temperatures resulted in lower histamine concentrations. For split noninjected fish, precooker temperature did not have a significant impact on histamine concentrations. No interaction was found between sampling time and any of the other factors. Thus, although some significant interactions were found between factors, no difference in the histamine inhibiting effect of precooking was found under this challenge study scenario.
The intent of this industrial precooking process validation study was to validate the use of a precooker EPIPT CL to control formation of histamine after precooking. This study was conducted using fish with elevated histamine concentrations and odors of decomposition. Previous research indicated that 60°C was an appropriate target maximum EPIPT for this study. Precooking to this target is a conservative CL and was used in this challenge study with worst-case fish.
The fish used here had histamine concentrations that exceeded the federal guidelines by 3- to 10-fold and were thus considered beyond worst case. Although the design of the spoilage portion of the experiment was to stop the histamine formation at 50 ppm, because changes in bacterial growth and histamine concentrations are variable and rapid after 50 ppm of histamine is reached, the general spoilage, bacterial growth, and increase in histamine concentration cannot be stopped immediately (28, 31).
This validation challenge study was conducted with freshly caught tuna that were live when caught and histamine free. These fish were immediately chilled and stored in ice or RSW until they were unloaded so the starting histamine concentrations remained very low. It was important that the fish not be frozen because freezing could significantly reduce existing HFB populations. Therefore, the subsequent spoilage process started with unfrozen fish of excellent quality.
The water flow set up, temperature control, and air circulation during the spoilage phase ensured that the water temperature was in the range optimal for HFB growth and histamine formation. The fish core temperatures rose from safe temperatures of −2 to 0°C to ambient temperatures close to 32°C, which are in the optimal range for HFB growth and histamine formation (Fig. 2). The deliberate spoilage process simulated a worst-case scenario of time-temperature abuse that could occur on board a harvest vessel due to delays in chilling and/or freezing of the fish. Resultant histamine concentrations in the tuna exceeded the FDA defect action level of 50 ppm (Table S1) on each spoiled experimental fish tested by sampling at the end of spoilage or tested later as an uncooked control.
The spoilage run was successful for creating fish with histamine concentrations well in excess of 50 ppm. Although the histamine concentrations at the end of the spoilage period in Suva were 66 to 214 ppm, the histamine concentrations of fish sampled before precooking at the processing facility were 150 to 450 ppm. These histamine increases could have occurred at two processing steps, given that the spoilage period had already successfully achieved a high growth rate of HFB and histamine production could not be immediately arrested (28, 31): (i) during the 1 h it took to load the fish into the blast freezer and for the fish to be frozen solidly (fish remained in the blast freezer for at least 24 h) and (ii) during the thawing phase (11 h in ∼26 to 28°C seawater) because the HFB present were not destroyed by freezing, transport, and storage. When thawed, the test fish (those to be precooked and the uncooked controls) also had strong odors of decomposition consistent with internationally used evaluation standards (36). Thus, this validation study was conducted with beyond worst-case frozen raw material that would not normally be accepted by tuna processors based on the CLs and sample sizes suggested by the FDA HACCP guidance (42).
These fish with very high histamine concentrations were precooked as described, and after precooking the fish core temperatures were reduced to 45°C in the sidespray room and further to 40°C in the conditioning room. The tuna only spent between 2 and 4 h in the cooling zones to lower the core temperatures, in contrast to general practices where time in the cooling area is maximized and exposure to ambient temperatures is minimized. These precooked high-histamine decomposing fish were then further challenged by exposure for up to 30 h to temperatures of 20 to 40°C, which are optimal for histamine formation (15).
The mean histamine concentrations and standard deviations for all the treatments and sample times are shown in Table 2. The large standard deviations indicate high variability in these histamine concentrations, which is consistent with the existing literature and is the reason for the lower FDA action histamine concentration of 50 ppm (31, 37).
In these experiments, precooking effectively stopped histamine formation for up to 18 h after precooking in these fish even though they had high initial histamine-forming capacity before precooking, as indicated by the raw fish control data. This histamine-forming capacity is a combination of both HFB and HDC, so both of these must have been destroyed and/or denatured during precooking. DeBeer et al. (9) found how much log lethality for HFB is accumulated during extra minutes at core temperatures of 56 or 57°C after precooking. According to Kanki et al. (20), 99% of the HDC is inactivated at 60°C. The target goal of this project was to precook the fish core to no more than 60°C. Because the temperature at the core continues to increase for some time after the steam is turned off (8, 29), a core temperature of 60°C means that the remaining portions of the fish loin and red meat are also at higher temperatures, effectively reducing the HFB levels or the HDC activity, thus stopping the formation of histamine until HFB growth resumes.
The large difference in histamine concentration between the raw controls and the precooked fish (df = 1, F = 198.06, P = 0.000) (Table S4) also indicates that the short duration of frozen storage did not affect the HFB and HDC (2) in the uncooked control fish. There was some concern that the freezing required to stabilize the tuna during transportation from Suva to the processing facility in Levuka would destroy the HFB, leading to protective effects and, thus, compromising the study. Based on a limited number of samples, Baranowski et al. (2) found that prolonged frozen storage for 24 to 40 weeks was responsible for a reduction in levels of surviving HFB in mahi-mahi (Coryphaena hippurus). However, in the present study, the HFB in the raw fish controls were not destroyed and were able to produce high concentrations of histamine in the absence of the precooking heat treatment; within 12 h, the concentration increased two- to sixfold (Table 2). The short period of frozen storage of less than 2 weeks in this study did not curtail subsequent histamine formation.
The results of this study indicate that histamine formation is delayed by precooking within the normal ranges of steam temperatures in precookers, 70 to 100°C. Consistent and significant differences in histamine concentrations were found between the raw fish controls and the precooked fish treatments just after precooking and for a significant period of time thereafter. These findings validate the efficacy of precooking to control HFB levels and HDC activity and, thus, curtail additional histamine formation and accumulation for 12 to 18 h.
This precooking validation study was conducted with fish that were beyond worst case: the minimum histamine concentration recorded after thawing and before precooking was 150 ppm (Fig. 2). No additional histamine formed until 18 h after precooking. The fish accepted by the tuna processors have a maximum histamine level of 30 ppm, which is five times lower than the minimum found in these spoiled fish. Therefore, these results support a CL of 12 h after precooking to control the formation of histamine with a significant margin of safety (6 h) for deviation evaluations (39). If a processor were to exceed the 12-h suggested time limit, a prudent corrective action based on chapter 7 of the HACCP guidance (42) could be used to determine the final disposition of the fish. EPIPT sampling plans, minimum sample sizes, and sampling strategies have been published previously (10, 11).
The following conservative CLs, as part of a HACCP plan built on a foundation of proper current good manufacturing practices, sanitation standard operating procedures, and prerequisite programs, will effectively control the hazard of histamine during the production of frozen tuna loins or canned products. These CLs are (i) accepting fish with less than 30 ppm of histamine and meeting sensory standards, (ii) controlling to less than 12 h the time elapsed for processing the fish from the start of thawing to the time precooker steam is turned on, (iii) precooking the fish to a core temperature of at least 60°C as measured after fish are removed from the precooker (25), and (iv) controlling to less than 12 h the time elapsed from the end of precooking to the start of retorting cans or freezing loins.
In conclusion, this experiment validated the process used during the production of canned tuna or frozen tuna loins of precooking the tuna to minimum core temperatures of 60°C to suppress the formation of histamine for 12 h or longer after precooking.
This work was fully supported and funded by Bumble Bee Seafoods and the Pacific Fishing Company. At the time these experiments were conducted, F. Adams (née Munshi, formerly Vogl) and F. Nolte were employees of Bumble Bee Seafoods and designed the experiments. F. Adams was the primary scientist on the ground in Fiji. Roque Salazar, Gerald Kontoh, Wayne Adams, and the plant operations team are gratefully acknowledged for their assistance in carrying out the research work in Fiji. Steven Mavity and Gina Ybanez (Bumble Bee Seafoods) are thanked for their critical reading of experimental design, helpful discussions, and insight. Scientists and statisticians from the U.S. FDA are acknowledged for their substantive contributions into the study design. J. Colton (at Minitab at the time of this research) coached the team on statistical design and data analysis. Dr. Steve Otwell (retired professor, University of Florida, Gainesville) is acknowledged for his subject matter expertise. Dr. Mona Baumgartel is thanked for her editorial assistance. L. Weddig and J. DeBeer are members of the National Fisheries Institute Canned Tuna Technical Committee and facilitated the publication of this article. The authors thank the journal reviewers for their very helpful suggestions to improve this manuscript. The preliminary results of this body of research were presented at a fisheries conference in 2012 (44) and have been used to validate the precooking CCP in place in various canneries exporting canned tuna to the United States.
Supplemental material associated with this article can be found online at: https://doi.org/10.4315/0362-028X.JFP-17-276.s1.