— OPINION —

Soft fruits like strawberries, raspberries, blackberries, and berry blends play an important part of an overall healthy diet. While millions of servings are consumed safely every day, they continue to be occasionally linked to hepatitis A (HAV) and norovirus outbreaks around the globe.

Investigation of a recent United States’ HAV outbreak, epidemiologically linked to frozen organic strawberries, has revealed complexities in assessing and determining the safety of frozen berry supply chains. These events have led to a recent call for HAV testing of finished products as a mandatory component by a retail food safety management program. We the authors describe the limitations of testing for HAV contamination in soft fruits, advise of the unintended consequences that could hinder more meaningful progress, and offer some alternative food safety management suggestions.

Limitations of the Enteric Virus test method
An important difference between testing for bacteria (e.g., Salmonella or L. monocytogenes) vs. viral pathogens in foods is that the viruses do not reproduce outside their host. This means culture enrichment is impossible, so testing protocols must rely on concentrating the virus from the food sample, followed by nucleic acid (RNA) extraction, and then detection of the virus genome using RT-qPCR. While this three-step method is currently the best approach available, it has inherent flaws. RT-qPCR detection limits tend to be higher than what we see for bacteria because of poor virus extraction efficiency and the presence of residual matrix-associated inhibitors. RT-qPCR is designed to detect only a small fragment of the viral genome. Since nucleic acids can be environmentally stable, detectable long after viruses are no longer infectious, a positive PCR result does not necessarily correspond to the presence of an intact and infectious viral particle. In short, a PCR positive cannot be automatically construed as an indication that the product contains infectious virus and constitutes a public health risk.

Limitations of sample size and sampling plans
HAV enters the berry supply chain through infected individuals, often via poor water and waste management systems. Unfortunately, the ultimate source of contamination is rarely identified through traceback, and the nature of any contamination event can vary widely from highly focal to much more diffuse. In many instances, HAV would be expected to be distributed in a non-uniform manner throughout a given lot of food. The table below was extracted from a simplified modeling exercise intended to investigate the likelihood of detecting HAV under different contamination scenarios¹. 

The analysis shows that the probability of detection is higher under what one might call a “gross” contamination event in which the entire production lot is contaminated. Note that this result is very dependent on the assumed concentration. If this assumed concentration is at or below the limit of detection for the test, even this contamination may not be detected.

Also, note that the likelihood of detection is always quite low when the contaminant enters the food through more focal events like poor personal hygiene of a hand harvester. Both concentration and prevalence of virus contamination drive the likelihood of detection. Detection probability can be improved by using bigger samples or testing more samples. Since sample size is fixed (e.g., 25 g for the ISO 15216-2 method), and virus testing costs usually exceed $150/sample, there is little opportunity for improving the utility of testing. In summary, sampling considerations, such as lack of statistical power, provide an additional hurdle to the value of screening finished products for HAV contamination. 

Table 1. Synopsis (best and worst case scenarios) of modeling exercise to investigate the likelihood of detecting HAV under different contamination scenarios with different numbers of samples (1 to 5) per lot tested using ISO 15216-2.

Nature of Contamination	Setting of Contamination	Distribution and Prevalence of Contamination	Concentration of Virus Particles in a Contaminated Unit	Probability of detecting viral RNA when using the ISO 15216 -2 method based on interpretation of results from one (low) to 5 (high) samples**
A. Poor personal hygiene; infected worker	Harvest	Focal or Non-uniform distribution (Assume approx. 0.075% of the lot is contaminated)	Assume high concentration (10-10,000 viral particles/g)	0.1 – 0.7%
B. Raw sewage entrains irrigation supply	Pre-harvest	Diffuse or Uniform distribution (Assume 100% of the lot is contaminated)	Assume low concentration (0.4 viral particles/g)	25.1 – 73.7%*

* This analysis relied on a number of assumptions: Virus may be present in some or all of a lot, and in different concentrations as detailed below; one virus particle is equivalent to one genome copy; when less than 1 virus particle is present in the ISO analytical unit of 1.25 g, the virus is present a fraction of the time (e.g., 0.1 virus particles per 1.25 g means the virus is present 10 percent of the time); This analysis also assumes that virus concentrations below the LOD₅₀ lower bound will always detected if present. This means that the high probability of detection from the raw sewage scenario may significantly over-estimate effectiveness.

** This high probability of detection is very sensitive to whether the virus concentration is above or below the assay limit of detection (LoD). If below the LoD, this probability is considerably lower.

Relationship between positive test result and public health risk
Taken alone, the RT-qPCR detection assay is highly sensitive, with a theoretical detection limit of one single target template molecule. However, this method has no reliable means by which to determine if that target template (1) comes from an infectious virus that can make someone sick; or (2) is just stray viral nucleic acid material or is associated with a non-infectious particle. In the first instance, there is a clear risk to public health; in the latter, essentially no risk. Until methods are available to determine whether a positive result is indicative of the presence of infectious virus (methods that are likely a decade or more away), interpretation of a positive PCR test is ambiguous. Interestingly, the Food and Agriculture Organization of the World Health Organization (FAO-WHO) has initiated a project to quantify the public health risk associated with potential enteric virus contamination in foods. Until an exercise like this is completed, the public health implications of testing and finding nucleic acid of HAV in frozen berries via random testing (especially when its concentration is very low), and without corresponding epidemiologically-linked illnesses is unclear. 

Where to go from here – addressing research gaps and a focus on prevention
The discussion above illustrates the myriad issues associated with detection of any non-cultivable foodborne pathogen in complex sample matrices. While testing is an important tool in the food safety toolbox, at the current time, there remain many research gaps that need to be answered to determine how to best use testing as a validation or verification tool. For example, could testing be used in a more focused and strategic manner as part of a risk assessment to determine potential sources of contamination such as irrigation water or wellness of a crew through testing portable toilet waste? Because HAV is so resistant to inactivation and disinfection, preventing contamination should be the top priority. Vaccination of food workers in direct contact with product along the supply chain is one means of minimizing the potential for contamination. While vaccination is expensive, so is testing, and a cost benefit analysis of farm crew vaccination should also be considered. A laser focus on personal hygiene practices of food handlers, Good Agricultural Practices, and proper water and waste management will all reduce the chance of an HAV contamination event and remain the current primary defenses for virus control in the berry supply chain. 

In summary, while tens of millions of fresh and frozen berries are consumed safely every day, unfortunately, berries are still occasionally implicated in HAV outbreaks and there is more the industry and regulators can do. Therefore, it is our view that an over-focus or reliance on end-product testing, using current vetted methods and low powered sampling plans, while well intentioned, could actually provide a false sense of security and hinder more meaningful and needed progress to further strengthen the safety of these important and nutritious products.

 Authors: Dr. Lee-Ann Jaykus, William Neal Reynolds Distinguished Professor, NC State University; Dr. Donald Schaffner, Graduate Program Director, Distinguished Professor and Extension Specialist, Rutgers University; Frank Yiannas, Former Deputy Commissioner, Food Policy and Response, U.S. Food and Drug Administration; Dr. Sanjay Gummalla, Senior Vice President of Scientific Affairs, American Frozen Food Institute

(To sign up for a free subscription to Food Safety News, click here.)