Antiviral Therapy Effectiveness and Threat Signal Composition
The safe mode threat signal \(\mathcal{T} = w_\text{cyto} dot C_\text{pro} + w_\text{ROS} dot [\text{ROS}] + w_\text{LPS} dot [\text{LPS}] + w_V dot V\) provides a quantitative framework for understanding why antiviral therapy produces inconsistent results across ME/CFS trials.
Antiviral therapy reduces the \(w_V dot V\) component of \(\mathcal{T}\). This disengages safe mode only if the viral load term is the dominant contributor to \(\mathcal{T}\)—that is, in early disease where cytokine and ROS contributions have not yet escalated. In late disease, \(w_\text{cyto} dot C_\text{pro}\) and \(w_\text{ROS} dot [\text{ROS}]\) dominate \(\mathcal{T}\), and eliminating \(V\) entirely reduces \(\mathcal{T}\) by only \(\approx w_V / (w_\text{cyto} + w_\text{ROS} + w_\text{LPS} + w_V) \approx 20%\)—insufficient to cross the safe mode disengagement threshold.
This explains the divergent results of antiviral trials. Lerner et al. showed benefit from long-term valacyclovir in EBV-positive patients (Lerner et al. 2007) (Lerner et al. 2002), while Montoya et al. found that valganciclovir benefit was confined to a subgroup with high baseline viral titers (Montoya et al. 2013). The model predicts that the responding subgroups are those with high \(w_V dot V / \mathcal{T}\) ratios—where viral load is a disproportionate contributor to the total threat signal. This is consistent with Hornig et al.’s finding that cytokine profiles in ME/CFS differ markedly between short-duration (\(< 3\) years) and long-duration (\(> 3\) years) disease (Hornig et al. 2015), suggesting that the composition of \(\mathcal{T}\) shifts over time as the model predicts.
Antiviral therapy effectiveness in ME/CFS is determined by the ratio \(w_V dot V / \mathcal{T}\)—the fraction of the total threat signal attributable to viral reactivation. Patients with high viral fraction (early disease, documented active viral replication, low cytokine burden) respond well; patients with low viral fraction (late disease, cytokine-dominant threat signal) do not, even if they carry the same reactivated viruses.
Certainty: 0.40. The prediction is directly derivable from the safe mode equation and explains existing trial data without requiring any new assumptions. The Lerner and Montoya findings (Lerner et al. 2007) (Montoya et al. 2013) are consistent with this interpretation.
Testable predictions:
- Antiviral response rate correlates inversely with disease duration (early \(>\) late), because disease duration correlates with the shift from viral-dominant to cytokine-dominant \(\mathcal{T}\).
- Antiviral response correlates with the ratio of viral reactivation markers to inflammatory markers at baseline (measurable as e.g. viral DNA copies / (IL-6 + CRP)).
- Combining antivirals with anti-inflammatory therapy (reducing \(w_\text{cyto} dot C_\text{pro}\) simultaneously) shows synergistic benefit exceeding either agent alone, because the combined reduction in \(\mathcal{T}\) may cross the disengagement threshold even when neither alone would.
- Patients with documented active viral replication (positive viral PCR) respond better than patients with only serological evidence (elevated IgG titers without active replication).
Limitations: Existing antiviral trials were not designed to measure \(\mathcal{T}\) components or stratify by viral fraction. Retrospective analysis of the Montoya and Lerner data could partially test the disease-duration prediction. The \(w\) weights are theoretical and have not been independently measured.