Genomic Surveillance and Epidemiologic Integration of the 2018-2020 Nord Kivu Ebola Virus Outbreak

Overview of the Tenth Ebola Virus Disease (EVD) Outbreak in the DRC

Outbreak Overview: On $1\,August\,2018$ , the Democratic Republic of the Congo (DRC) declared its tenth EVD outbreak, primarily affecting the Nord Kivu and Ituri provinces.
Outbreak Duration: The Nord Kivu EVD outbreak transitioned from $1\,August\,2018$ to its conclusion on $25\,June\,2020$ .
Genomic Surveillance Effort: The Institut National de Recherche Biomédicale (INRB) implemented an end-to-end genomic surveillance system to aid the epidemiologic response.
Sample Coverage: * Researchers generated $792$ full and partial genome sequences. * $744$ of these were new genomes sampled between $27\,July\,2018$ and $27\,April\,2020$ . * $48$ genomes were previously available. * This collection represents approximately $24\%$ of all laboratory-confirmed EVD infections in the DRC during the analyzed period.
Historical Context: Since the first outbreak in Yambuku in $1976$ , further outbreaks have occurred sporadically. In $June\,2018$ , whole-genome EBOV (Ebola virus) sequencing capacity was established at INRB in Kinshasa, enabling this real-time surveillance.

The End-to-End Genomic Surveillance System

System Components: The system included sequencing, bioinformatics analysis, and dissemination of genomic epidemiologic results to frontline public health workers.
Evolution of Communication: * Phase 1 (Initial): Genomic findings were communicated through haplotype maps, manually annotated with epidemiologic info and shared as PDFs. * Phase 2 (September 2019 Transition): Switched to an automated pipeline creating interactive situation reports known as Nextstrain Narratives. These reports were available in English and French and allowed self-guided data exploration.
Response Timing and Lag: * The average time between sequencing and private sharing with frontline teams was $6.6\,days$ (standard deviation = $7.8\,days$ ) after the automated pipeline was implemented. * Public release of data followed on average $13.4\,days$ later. * Under ideal circumstances, the window from diagnostic testing to delivering inferences took as little as $7\,days$ ( $4\,days$ for testing/prep and $2-3\,days$ for sequencing/analysis). * Success Metrics: Before $1\,September\,2019$ , $33\%$ ( $169$ of $508$ ) of samples were analyzed within $30\,days$ of collection. After that date, the proportion rose to $48\%$ ( $128$ of $264$ ).
Data Sharing: Deidentified metadata and sequences were released publicly on GitHub (https://github.com/inrb-drc/ebola-nord-kivu).

Broad-Scale Spatiotemporal Dynamics of EVD Circulation

Zoonotic Origin: Phylogeographic analysis inferred a single zoonotic spillover event occurring around $July\,2018$ in the Mabalako health zone.
Clade Identification: * Primary Outbreak Clade: Defined by the mutation $A7312G$ . It emerged from an introduction from Mabalako into Beni in $August\,2018$ ( $95\%$ Confidence Interval: $15-20\,August\,2018$ ) and became the primary lineage. * Secondary Outbreak Clade: Resulted from an introduction from Beni into Katwa between $August$ and $October\,2018$ . This lineage persisted in Mandima and Rwampara until at least $September\,2019$ .
Migration Patterns: * Introduction Volume: Researchers detected $188$ independent introduction events with at least $80\%$ confidence. * Key Health Zones: Five health zones acted as the primary sources for seeding transmission elsewhere: Beni, Mabalako, Katwa, Kalunguta, and Mandima. Each seeded at least $20$ separate instances of transmission into other zones. * Transmission Distance: $50\%$ of movement events occurred between health zones less than $49\,km$ apart; $95\%$ were less than $200\,km$ apart. * Duration of Local Circulation: Most lineages lived briefly in a specific health zone. $50\%$ circulated for less than $10\,weeks$ ; $95\%$ circulated for less than $40\,weeks$ . * Regional Dynamics: The frequent movement of lineages with short-lived local transmission chains mirrors the dynamics of the $2013-2016$ West African EVD outbreak.

Case Study 1: Superspreading and Vaccine Allocation

Vaccination Strategy: Due to limited supplies of rVSV-ZEBOV-GP and Ad26-ZEBOV/MVA-BN-FILO vaccines, efforts initially focused on contacts and contacts-of-contacts of confirmed cases.
Clergy as Superspreaders: * Case KAT5915: A pastor who died of EVD in Beni. His funeral in Butembo, conducted without safe burial protocols, led to cases in Beni, Butembo, Ariwara, and Oicha. * Impact: $320$ sequenced infections were descended from this single founder event.
Taxi-Drivers as Vectors: * Case MAN12309: A motorcycle taxi driver who worked while symptomatic in $December\,2019$ . * Impact: $20$ contacts had identical EBOV sequences to his, confirming him as the likely source.
Policy Outcome: Vaccination recommendations were expanded to include preemptive vaccination for clergy and motorcycle taxi drivers.

Case Study 2: Differentiating Reinfection from Relapse

Clinical Scenario: In $December\,2019$ , a male patient previously infected and recovered in $June\,2019$ presented with EVD symptoms again.
The Question: Was this a new infection (reinfection) or a reactivation of the original virus (relapse)?
Genomic Evidence: * The patient’s $December$ specimen was genomicly similar to viral lineages in Mabalako from $June\,2019$ rather than those circulating in $December\,2019$ . * Comparison between samples MAN4194 ( $June$ ) and MAN12309 ( $December$ ) showed only $2$ nucleotide differences ( $T5578C$ and $A6867G$ ). * This is far fewer substitutions than expected based on the outbreak substitution rate of $1.17 \times 10^{-3} \text{ substitutions per site per year}$ .
Conclusion: The data supported a scenario of relapse after treatment with monoclonal antibodies, rather than reinfection. This finding was critical for regulators evaluating interventions.

Operational Challenges and Logistical Constraints

Geography: Sequencing was initially restricted to Kinshasa, $2,600\,km$ away from the outbreak. Specimens had to transit through regional labs to Beni, then Goma ( $240\,km$ ), and finally to Kinshasa.
Transport Issues: Commercial airlines initially refused to carry EBOV-positive specimens. While WHO eventually arranged flights, this contributed to significant lag.
Laboratory Conflict and Safety: * A mobile lab in Katwa was established in $February\,2019$ to mitigate lag, but faced extreme security risks. * The lab was located next to an Ebola Treatment Unit destroyed by arson. Access was often banned or limited to $2-hour$ windows with armed escorts, which was insufficient for full sequencing protocols.
Infrastructure and Supplies: * Equipment like gloveboxes had to be shared between diagnostic and sequencing teams, with diagnostics prioritized. * Internet outages, including a $3-week$ shut-off during the $January\,2019$ federal election, hindered remote technical support. * Reagents were frequently hand-carried into the country because standard shipping led to customs delays where reagents would thaw and degrade.

Integration and Methodology

Methodological Details: * Diagnostic Testing: GeneXpert Ebola Assay (Cepheid). * Sequencing Methods: Hybrid capture method (using KAPA RNA HyperPrep and Illumina iSeq/MiSeq) and amplicon-based method (using PrimalSeq and Nextera DNA Flex). * Bioinformatics Tools: Augur (alignment/processing), MAFFT (sequence alignment), IQ-TREE (maximum likelihood phylogeny), and TreeTime (temporal resolution and discrete trait inference).
Lessons Learned: * The study highlights the need for integrated databases where epidemiologic, laboratory, and genomic data are linked. * Sustaining in-country capacity is vital; the system built for Ebola in the DRC is now being utilized to track SARS-CoV-2. * Genomic surveillance during the Nord Kivu outbreak covered $24\%$ of cases, a significant increase from the $5\%$ coverage during the $2013-2016$ West African epidemic.