Our findings characterise patterns of HSV‐2 reactivation and quantify rates of change that are important for understanding transmission and guiding clinical management. We found that among this group of women in the early years of infection, HSV‐2 reactivation rates varied >tenfold. The duration of HSV‐2 mucosal shedding associated with each distinct reactivation was typically 1–3 days, and this duration varied little among the individuals in the study. Thus, most of the variation in HSV‐2 reactivation patterns was due to differences in frequency rather than duration of viral reactivation. Our results further suggested that high frequency of reactivation could lead to overlapping of reactivations, with extended mucosal shedding episodes composed of two or more reactivations. The frequency of reactivation and recurrences generally waned over time, but some individuals continued to experience high rates 4–5 years after primary infection.
- Viral dynamic modelling of clinical data suggests that single episodes of HSV‐2 mucosal shedding could involve multiple reactivations in the ganglia. Consequently, rates of HSV‐2 reactivation could be higher than suggested by mucosal shedding rates.
- The frequency of HSV‐2 reactivation varies widely among individuals and decreases over time; in contrast, the duration of mucosal shedding associated with a reactivation varies little and does not appear to decrease over time.
- Modelling methods can provide insight into viral pathogenesis and suggest novel approaches for experimental biology.
Our use of the model of HSV within‐host dynamics provided several benefits. First, the model provided estimates of the frequency and duration of episodes of mucosal shedding, which are uncertain when mucosal specimens are collected at spaced time points. Second, the model provided estimates of the rate of HSV reactivation in ganglia, which is not feasible to measure directly. In order to obtain these estimates, the model assumes that individuals continue to experience reactivations of HSV from latency at their characteristic rate regardless of whether they are currently shedding HSV from the mucosa. The validity of this assumption is difficult to confirm with laboratory or clinical studies. However, our model validation showed that mucosal shedding data computer‐simulated using this assumption closely matched observed mucosal shedding patterns, supporting the plausibility of the model. Studies are planned to further evaluate the model.
The modelling results suggest that reactivations occur at a higher frequency but shorter duration than mucosal shedding episodes, with the implication that some episodes of mucosal shedding are a composite of two or more distinct reactivation events. This has several implications. It suggests that true rates of HSV reactivation from latency could be higher than previously surmised based on the frequency of mucosal shedding episodes. This in turn implies that the stimuli that trigger reactivation could be more numerous and diverse than currently appreciated. These stimuli are still incompletely characterised. Because we are able to quantify reactivation rates, our methods could be used to better characterise the determinants of reactivation frequency and the effect of antiviral therapy on HSV‐2 shedding.
Identifying frequent reactivators and targeting them for antiviral therapy can be an effective public health strategy for reducing transmission of HSV‐2.15
Here we have quantified biological heterogeneity in reactivation frequency and the rate of waning over time. This information can be used to identify the virological core groups that are disproportionately responsible for transmission, leading to more effective control.
Our study is the first to apply viral dynamic modelling to longitudinal data on HSV infection. Applying these techniques to our sample of 18 women, the largest repeated longitudinal data set currently available, is sufficient to demonstrate the feasibility and benefits of these new methods and provide insights into typical HSV reactivation rates and patterns. Future studies with larger and more diverse samples, including men, will allow inferences about a wider population of infected patients.
Our viral dynamic model decomposes HSV pathogenesis into two distinct processes, reactivation of HSV in the ganglia and subsequent shedding from the mucosa, and provides distinct estimates of these processes for each individual. Because of this approach, our methods can be used to evaluate the differential efficacy of antiviral therapies on the two processes. This approach could facilitate the development of treatments that are targeted to a specific process or compartment.