Uncovering the viral aetiology of undiagnosed acute febrile illness in Uganda using metagenomic sequencing

Abstract

Viruses associated with acute febrile illness in Africa cause a spectrum of clinical disease from mild to life-threatening. Routine diagnostic methods are insufficient to identify all viral pathogens in this region. In this study, 1281 febrile Ugandan patients were prospectively recruited as part of the CDC-UVRI Acute Febrile Illness Study and pre-screened for common pathogens. 210/1281 undiagnosed samples, and 20 additional samples from viral outbreaks were subjected to metagenomic sequencing. Viral pathogens were identified in 44/230 (19%), including respiratory, hepatitis, blood-borne, gastrointestinal and vector-borne viruses. Importantly, one case of Crimean-Congo haemorrhagic fever and two cases each of Rift Valley fever, dengue and yellow fever were detected in 7/230 (3%) of cases. Le Dantec virus, last reported in 1969, was also identified in one patient. The presence of high-consequence and (re-)emerging viruses of public health concern highlights the need for enhanced population-based diagnostic surveillance in the African region.

Fig. 1. Undiagnosed viruses identified in the AFI study.

a Map showing AFI study sites in Uganda and viruses identified from undiagnosed samples listed by site (red star). Approximate locations of outbreak samples, where available, are also marked (blue star). b Viruses detected by mNGS grouped as zoonoses/arboviruses (purple), hepatitis (yellow), blood-borne infections (pink), respiratory viruses (blue), gastrointestinal infections (brown), viruses associated with rash (mauve) and uncharacterised (grey). Icons show the number of patients infected with each virus. RVFV rift valley fever virus, HAV hepatitis A virus, CCHFV Crimean-Congo haemorrhagic fever virus, HEV hepatitis E virus, HHV6b human herpesvirus-6b, HIV human immunodeficiency virus, LDV Le Dantec virus, HBV hepatitis B virus, HPIV-3 human parainfluenza virus 3, SAFV Saffold virus. Icons to the right indicate syndromic grouping/transmission routes. Species silhouettes obtained through Phylopic; tick, rodent, cattle https://creativecommons.org/publicdomain/zero/1.0/, mosquito https://creativecommons.org/publicdomain/mark/1.0/, primate (credit: Bogdan Bocianowski) https://creativecommons.org/licenses/by-sa/3.0/.

Fig. 2. Viruses associated with haemorrhagic fever.

a–c Coverage plots for CCHFV, YFV, DENV identified by mNGS showing number of mapped reads (y-axis) at each nucleotide position across the genome (x-axis) for each virus. Annotated reference Orthonairovirus and Flavivirus genomes indicate regions covered by sequencing reads. Genes are shown as polymerase (L) and non-structural proteins in blue, capsid and nucleoprotein in pink and surface proteins in yellow. d Phylogenetic tree of the full-length nucleotide sequence of CCHFV showing placement of patient isolate for L, M and S segments. Tip labels show the strain name and country of isolation. Colours indicate geographically distributed lineages of CCHFV. e Full genome nucleotide tree of DENV genotypes showing country and year of isolation. f Full genome nucleotide tree of YFV sequences showing country and year of isolation. The evolutionary scale is shown below for each tree. Ultrafast bootstrap values >90 are shown for nodes. Patient samples are highlighted in grey. Substitution models used: GTR + F + I + G4 (CCHFV-L), GTR + F + R3 (CCHFV-M), TIM2 + F + I + G4 (CCHFV-S, DENV) and GTR + F + R2 (YFV).

Fig. 3. Le Dantec virus (LDV).

a Coverage plots for LDV identified by mNGS showing number of mapped reads (y-axis) at each nucleotide position across the genome (x-axis) for each virus. The annotated reference genome below plot shows LDV genes: nucleoprotein (N), phosphoprotein (P), matrix (M), glycoprotein (G), accessory gene (U1), polymerase (L). b ELISA analysis showing IgG antibody response to recombinant LDV glycoprotein in patient 220-2, contact of the patient and a healthy individual from Uganda. The plasmid containing an Ephrin β2 insert was used as a mock recombinant protein. The plot shows mean absorbance OD values at 450 nm (y-axis) and standard deviation (error bars) against increasing serum dilutions (x-axis) for n = 3 technical replicates from one representative experiment. c Pseudotype neutralisation assay showing luciferase reporter activity (y-axis) against increasing serum dilutions (x-axis) for patient 220-2, contact and healthy serum using LDV pseudotype virus compared with VSV control. Data shows the mean and standard error of mean for n = 3 technical replicates from one representative experiment. d Full genome maximum-likelihood phylogenetic nucleotide tree using the GTR + F + R4 model for Ledantevirus sequences illustrating the reported host for each virus. Evolutionary scale and ultrafast bootstrap values >90 are shown. The patient sample is highlighted in grey. Species silhouettes obtained through Phylopic; tick, midge, boar, bat fly, rodent, louse fly https://creativecommons.org/publicdomain/zero/1.0/, mosquito https://creativecommons.org/publicdomain/mark/1.0/.

Fig. 4. Other viral pathogens identified from patient samples.

Coverage plots showing number of mapped reads (y-axis) against genome position (x-axis) for each patient, for a HAV, b HEV, c Enteroviruses including Echoviruses, Coxsackievirus and Rhinovirus C, d Measles virus; and e HIV-1. Reference genome drawings indicate depth for different regions along the sequence. Annotated reference genomes show gene positions. Structural genes/surface proteins are shown in yellow, and other genes in blue, brown, purple, pink and grey.

Fig. 5. Clinical and environmental variables associated with viral infection.

a Multivariable logistic regression analysis (two-sided) of factors associated with viral infection for n = 1281 patients. Environmental and clinical history variables reaching p = <0.05 were included in a stepwise analysis. 95% confidence intervals are shown as error bars. Factors found to reach statistical significance are shown in black, while those not reaching significance are shown in grey. b Diagnoses associated with haemorrhage and fever. Cumulative totals of haemorrhagic symptoms in relation to the underlying diagnosis. Presentation with the viral haemorrhagic fever viruses CCHFV, RVFV and YFV was not associated with haemorrhage in the AFI cohort. c Initial clinical diagnosis in cases of confirmed viral infection. d Antimicrobial therapies administered in cases of confirmed viral infection.