Stability of Avian Influenza A(H5N1) Virus in Milk from Infected Cows and Virus-Spiked Milk

TO THE EDITOR:

In late March of 2024, federal and state veterinary and public health agencies in the United States (US) reported the first outbreak of highly pathogenic avian influenza A(H5N1) (HPAI H5N1) virus in dairy cattle, with virus detected in clinical milk samples from symptomatic animals¹. Viral genetic fragments were subsequently found in pasteurized retail dairy products², raising concerns about potential exposure through the human food supply. Heat treatment has been shown to reduce or eliminate infectious viruses from both infected cows’ milk³ and milk spiked with HPAI H5N1 virus⁴. However, it is unclear whether HPAI H5N1 viruses in spiked milk replicate the properties of viruses in infected cows’ milk. We therefore directly compared the stability of HPAI H5N1 viruses (Table S1; for details, see the Supplementary Appendix, available with the full text of this letter at NEJM.org) in infected cows’ milk versus spiked milk at 4 °C (refrigeration temperature) and after heat treatment at 63 °C (low-temperature long-time pasteurization) or 72 °C (high temperature short time pasteurization).

We previously reported that HPAI H5N1 viruses in raw milk from infected cows stored at 4 °C remained infectious for at least 5 weeks, with a modest 2-log decline in titer³. An extended analysis of four unique raw milk samples from infected cows showed that infectious viruses could be detected for up to 22 weeks at 4 °C (Fig. 1a), with decay rates only slightly greater than those of virus-spiked medium or PBS (Fig. 1b). In contrast, the same viruses were undetectable within 2 weeks of being spiked into raw milk from healthy cows (Fig. 1a). Similar rapid decay (within 3 weeks) was observed for five other influenza viruses spiked into healthy cows’ milk, whereas stability was retained in PBS or medium (Fig. S1). These observations suggest enhanced stability of viruses produced in the mammary gland.

Figure 1. Stability of HPAI H5N1 viruses.

(a) and (b) Long term stability at 4 °C. Aliquots of infected cows’ milk (one replicate per milk sample) or virus-spiked raw milk, PBS, or MEM-BSA (three replicates per sample) were stored at 4 °C and titrated weekly by plaque assays in MDCK cells until infectious virus was no longer detected. In panel (a), each plot shows virus titers for one infected cows’ milk sample (indicated at the top of the plot: NM93, NM115, KSSM3, or KSSM6) and milk, PBS, or MEM-BSA spiked with the virus isolated from the same milk sample (see Table S1). The “†” symbol indicates that the sample volume was depleted before infectious virus dropped below the limit of detection. Panel (b) gives the decay rates for infectious viruses in each type of sample (note, the color code is the same as in panel (a)). Points represent the mean decay rate and error bars represent the standard deviation. PFU/ml, plaque forming units per milliliter; wk, weeks. (c) and (d) Stability after heat treatment. Aliquots of infected cows’ milk or virus-spiked raw milk were subjected to heat treatment and then tissue culture infectious dose 50 (TCID₅₀) half-lives were inferred using a Bayesian framework. Panel (c) shows TCID₅₀ decay kinetics of one replicate experiment for milk sample TX97–37 (left) or raw milk spiked with the virus isolated from the TX97–37 milk sample (right) treated at 63 °C (see Table S1). In each graph, orange points show the posterior mean TCID₅₀ values for replicates at each timepoint inferred independently. TCID₅₀ decay was measured in the posterior distribution of the time-dependent TCID₅₀ model, which is visualized here as straight lines. The posterior mean is shown as a white dashed line, and 200 posterior samples are shown as thin orange lines that capture the distribution of credible fits. The gray dashed line indicates the limit of detection in the TCID₅₀ assays. Panel (d) gives the Δhalf-life values for all infected cows’ milk versus spiked milk comparisons at 63 °C and 72 °C. Thin and thick horizontal lines show the 95% and 50% credible intervals, respectively. sec, seconds.

We previously showed that heat treatment of HPAI H5N1-infected cows’ milk reduced virus titers below detectable levels (63 °C, 5 min) or by ≥ 4.5 log units (72 °C, 30 sec)³. We now compared heat inactivation at 63 °C or 72 °C for milk from infected cows versus milk spiked with matched virus isolates. While untreated samples had similar titers (Fig. S2), viruses in spiked milk decayed more rapidly at 63 °C, with shorter half-lives in 2 of 3 samples (Fig. 1c—d, Fig. S3). Differences were less pronounced at 72 °C (Fig. 1d, Fig. S4). These data suggest that viruses produced in the mammary gland are more heat-stable than viruses spiked into milk, particularly at lower pasteurization temperatures.

These findings demonstrate that HPAI H5N1 virus is more stable in milk from infected cows than in milk spiked with matched virus isolates, both during cold storage and after heat treatment. The mechanisms are not known but may involve protective factors in milk such as fats, casein proteins, or debris. While pasteurization has been shown to inactivate HPAI H5N1 viruses, the persistence of infectious viruses in raw milk highlights the potential risk of transmission through unpasteurized dairy products, both for humans and domesticated animals. Overall, our results suggest that, whenever possible, materials from infected animals should be used to examine the characteristics of bovine HPAI H5N1 viruses in dairy products and the associated public health risks.