Background: Viral load in Ebola virus disease (EVD) correlates with case fatality rate (CFR) and is an important parameter used for diagnostic cutoffs, stratification in randomized controlled trials, and epidemiologic investigation. However, it is currently only estimated using cycle threshold (Ct) values from diagnostic qRT-PCR tests that were not developed or validated for quantitative purposes. Furthermore, during the 2013-2016 EVD epidemic in west Africa, numerous different diagnostic platforms and protocols were used and yielded inconsistent, often conflicting data. Methods: 727 whole blood samples, both positive and negative for Ebola virus (EBOV), that were collected for diagnostic purposes in Monrovia, Liberia between August 28 and December 18, 2014 were reanalyzed under highly controlled biosafety level 4 laboratory conditions. An optimized TRIzol-based extraction protocol was utilized, followed by a novel semi-strand specific two-step droplet digital PCR assay to quantify viral load as log10 EBOV RNA copies/mL. These data were then compared to the cycle threshold (Ct) values from the original diagnostic qRT-PCR testing, and a subset of the samples was also reanalyzed following re-extraction with the same diagnostic one-step PCR and instruments employed at the Ebola treatment unit for further assessment. Finally, for EBOV-positive patients, correlations between viral load, outcome, time from symptom onset, and other epidemiological parameters were evaluated utilizing a deidentified database. Findings: Survivors had significantly lower mean viral loads at presentation than non-survivors in both the reanalyzed dataset (5.61 [95% CI 5.34 to 5.87] versus 7.19 [95% CI 6.99 to 7.38] log10 EBOV RNA copies/mL, respectively; p < 0.0001) and diagnostic dataset (Ctvalue 28.72 [95% CI 27.97 to 29.47] versus 26.26 [95% CI 25.72 to 26.81], respectively; p < 0.0001). However, receiver operating characteristic curve analysis demonstrated the reanalyzed dataset to be superior to the diagnostic dataset (area under curve 0.80 [95% CI 0.75 to 0.85] versus 0.66 [95% CI 0.59 to 0.72], respectively) in distinguishing between survivors and non-survivors and the prognostic capacity of viral load increased substantially upon reanalysis (odds ratio [OR] of death 8.06, 95% CI 4.81 to 13.53, p < 0.0001 for viral loads above 6.71 log10 EBOV RNA copies/mL versus OR of death 2.02, 95% CI 1.27 to 3.20, p = 0.0028 for Ct values below 27.37). Diagnostic Ct values underestimated viral load in non-survivors by nearly 100-fold on average (mean difference between diagnostic Ct value and reanalyzed Ct value of 5.15 [95% CI 4.43 to 5.87]); substantially less discrepancy was observed for survivors (mean difference between diagnostic Ct value and reanalyzed Ct value of 1.79 [95% CI 1.16 to 2.43]). Six samples that were reported negative by diagnostic testing were found to be positive upon reanalysis, and had high viral loads. Non-survivors presented to the ETU significantly earlier as the epidemic progressed, from 6.8 days following symptom onset (95% CI 5.7 to 7.9), on average, during the first three weeks to 3.7 days (95% CI 2.4 to 5.0) during the last three weeks (p = 0.0006) and had significantly lower viral loads (7.6 log10 EBOV RNA copies/mL [95% CI 7.3 to 7.9] to 6.5 log10 EBOV RNA copies/mL [95% CI 5.8 to 7.2] for the same intervals (p = 0.0046). However, the CFR remained steady at approximately 60% throughout. Interpretation: Inaccurate viral load estimation from diagnostic Ct values is likely multifactorial, but unaddressed PCR inhibition from tissue damage in patients with fulminant EVD may largely account for the discrepancies between the datasets in this study. In light of these findings, testing protocols for EVD require further standardization and validation to produce accurate viral load estimates, minimize false negatives, and allow for reliable epidemiological investigation. Funding Information: This research was funded by the Intramural Research Program of the National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH). Declaration of Interests: The authors declare no conflicts of interest. Ethics Approval Statement: The deidentified samples and data utilized for this study received a determination of ‘not human subjects research’ by the NIH Office of Human Subjects Research Protection and has been approved by the University of Liberia-Pacific Institute for Research and Evaluation IRB.
Read full abstract