History and Epidemiology

On March 20, 2015, a 44-year-old woman from Montserrado County, Liberia, was confirmed to have EVD. Blood samples from the patient were confirmed to be positive for EBOV RNA by quantitative reverse-transcriptase–polymerase-chain-reaction (RT-PCR) assay at the Eternal Love Winning Africa (ELWA) Laboratory in Paynesville, Liberia. The patient died on March 27, 2015.

The case investigation did not reveal an immediate source of infection, such as contact with patients with acute EVD. However, the patient reported that on March 7, 2015, she had had unprotected vaginal intercourse with a male Liberian survivor of EVD.8 Subsequent to the patient’s EVD diagnosis, 192 contacts were identified,11 all of whom were free from clinical signs.

The survivor also lived in Montserrado County. Several members of his family had had EVD, beginning in late August 2014. The survivor’s older brother, who presented with clinical signs of EVD on August 22, died during the night on September 5–6, 2014, and was confirmed to be positive for EBOV RNA by means of a postmortem quantitative RT-PCR assay. The survivor is thought to have had symptoms of EVD beginning on September 9, 2014, which is the estimated triage date,8 and he was admitted to the nearby Island Clinic Ebola Virus Disease Treatment Unit on September 23.

Quantitative RT-PCR testing for EBOV RNA in the survivor’s first blood sample on September 28 yielded ambiguous results. Repeated testing of this sample on September 29 yielded a negative result for EBOV. A subsequent test performed on October 3 (presumably from a second blood sample, although this information could not be confirmed owing to the absence of a sample record) was also negative. The survivor was discharged from the Ebola treatment unit on October 7 and reported no subsequent illness.

On September 20, clinical signs of EVD developed in the survivor’s former wife, who was estranged from the survivor. She was admitted to an ELWA Ebola treatment unit on September 24 and died the following day.

As a result of the case investigation into the patient’s illness, the survivor voluntarily provided a blood sample on March 23, 2015, and a semen sample on March 27, 2015 (199 days after the estimated onset of EVD and 175 days after the survivor’s blood tested negative for EBOV). The blood sample tested negative for EBOV RNA on quantitative RT-PCR assay, but the sample tested positive for EBOV glycoprotein-specific and nucleoprotein-specific IgG antibodies.8 The semen sample tested positive for EBOV RNA on quantitative RT-PCR assay, but attempts to culture virus were unsuccessful.

Figure 1. Figure 1. Clinical Timelines for the Patient and the Survivor, from September 2014 through May 2015. Shown are key dates regarding the Ebola virus disease (EVD) presentation, diagnostic tests, and outcomes for the survivor (S) and the patient (P). Horizontal bars estimate the number of days of persistence of the Ebola virus (EBOV) since the date of disease onset and since the date of clearance from blood. ETU denotes Ebola treatment unit.

On April 28 (32 days later), the survivor provided a second semen sample for diagnostic testing at the Liberian National Public Health Reference Laboratory in Margibi County. No EBOV RNA was detected by the quantitative RT-PCR assay. A third semen sample, collected 3 days later, on May 1, also tested negative for EBOV on quantitative RT-PCR assay, which suggests that there was possible EBOV clearance from semen 231 days after the estimated onset of EVD and 207 days after the survivor’s blood tested negative. A timeline of the events is shown in Figure 1.

Specific consent was obtained from the survivor. For all other samples collected for testing during the EVD outbreak, informed consent was not obtained because this work was conducted at the Liberian Institute for Biomedical Research (LIBR) as part of the EVD response and EBOV surveillance. With the consent of the National Incident Management System of the Ebola Virus Disease Outbreak and the Liberian Ministry of Health and Social Welfare, the work was supervised by the LIBR institutional review board. All the information obtained from the participants was anonymized for this report. The WHO Liberia Country Office team coordinated field epidemiologic investigations and support to the survivor and the patient.

Molecular Investigation

As part of the investigation into the source of the patient’s EBOV infection, the following samples were examined: whole blood from the patient was tested on March 20 and 21, 2015 (two samples); whole blood from the survivor’s older brother was tested on September 9, 2014; whole blood from the survivor’s former wife was tested on September 24, 2014; and semen from the survivor was tested on March 27, 2015. Viral RNA that was potentially present in all five samples was initially sequenced on an Illumina MiSeq at the LIBR with the use of methods that have been described previously.12

Table 1. Table 1. Distinct Ebola Virus Genome Substitutions in the Patient, the Survivor, and the Survivor’s Older Brother.

Nearly complete EBOV genome sequences (97.4 to 99.7% coverage) were assembled from the samples obtained from the patient, the survivor’s older brother, and the survivor’s former wife. No EBOV sequences were obtained from the survivor’s semen sample with the use of this sequencing method. Therefore, we enriched the semen sample for EBOV genomic RNA using the TruSeq RNA Access kit (Illumina) with custom capture probes designed against EBOV, along with other modifications (see the Supplementary Appendix, available with the full text of this article at NEJM.org). The semen sample was processed and sequenced separately to avoid contamination. With the combined data from four independent enrichment libraries, 85.1% genome coverage was achieved. A minimum of 3× sequencing depth was required to determine a genome position. However, the enrichment process resulted in a large number of duplicate reads. Therefore, the duplicate-adjusted sequencing depth for the semen sample was less than 3× in some positions (Table 1). The assembled genomes are available at GenBank under accession numbers KT587343, KT587344, KT587346, and KT587345.

Figure 2. Figure 2. Median-Joining Haplotype Network. This network was constructed from a full genome alignment of 100 clinical sequences of the Ebola virus Makona variant, including those assembled from blood samples obtained from the patient (P), the survivor’s older brother (SB), and the survivor’s former wife (SFW), from a semen sample from the survivor (S), and from 96 additional genomes chosen from the 796 genomes that were analyzed to be representative of samples collected in Guinea, Liberia, Mali, and Sierra Leone. For visual clarity, the network was limited to 100 genomes. The GenBank accession numbers for the tested genomes are as follows: for P, the number is KT587343, for S, the number is KT587344, for SB, the number is KT587346, and for SFW, the number is KT587345. Each colored vertex represents a sampled viral haplotype. The vertex size is proportional to the number of sampled sequences. Genomes sequenced in this study are shown in pink. Purple vertexes (SPB) indicate samples from the last known cluster of EVD cases in Liberia before the infection of the patient discussed in this report. Other colors indicate the respective countries of origin. Edges are not drawn to scale; hatch marks indicate the number of substitutions along each edge. The vertex SL2 represents the ancestral haplotype that is thought to have been introduced into Liberia in the spring of 2014.12,15

The four assembled EBOV genomes were compared with all publicly available sequences (796 genomes, including 56 from cases in Liberia12-19) from the outbreak in Western Africa (Makona variant).20 The results were consistent with sexual transmission of EBOV from the survivor to the patient. First, the EBOV genome from the patient grouped phylogenetically with other genomes obtained from Liberian patients and was distinct from sequences from patients in Guinea, Sierra Leone, and Mali (Figure 2). Thus, it is unlikely that the patient was infected owing to an undocumented reintroduction of EBOV to Liberia from a neighboring country with ongoing transmission.

Second, before the confirmation of EVD in the patient, the last known cluster of EVD cases in Liberia (December 29, 2014, to February 19, 2015) was linked to a single index case from a village near Saint Paul River Bridge, and the three sequenced EBOV genomes from this cluster (LIBR0993, LIBR1195, and LIBR1413) grouped together in an evolutionary lineage (SPB in Figure 2) that was unrelated to the EBOV genome from the patient.21 Therefore, the infection in this patient is unlikely to have originated from this cluster of EVD cases.

Finally, the EBOV genomes from the patient and the survivor differed in only one position (11,263) across 15,808 nucleotides, a finding that is consistent with direct EBOV transmission (Figure 2 and Table 1). Notably, EBOV genomes from the patient and the survivor shared eight substitutions relative to the ancestral haplotype (SL2)15 that is thought to have been originally introduced into Liberia,12 and three of these substitutions have thus far been seen only in the viruses that infected the patient and the survivor (Figure 2 and Table 1). Although only 1× sequencing depth was obtained for several of these positions (after correction for duplicate reads), the detection in the survivor’s semen sample of every substitution that distinguished the patient’s EBOV sequence from the ancestral SL2 haplotype is indicative of a close epidemiologic link. The EBOV genome obtained from the survivor’s older brother shared five of eight substitutions with the EBOV genomes of the patient and the survivor, which suggests involvement of the survivor’s older brother in the same transmission chain (Figure 2 and Table 1). The EBOV genome from the survivor’s former wife, however, was distinct (Figure 2).

The EBOV genomes from the patient and the survivor differed at a single position (11,263), thus representing an additional substitution in the survivor’s EBOV genome relative to all other genomes assembled. After duplicates from PCR amplification were controlled for, this position had only 1× coverage depth. Given the low level of sequencing depth, this apparent substitution may simply represent a low-frequency allele in the survivor’s EBOV population or even a sequencing artifact. Alternatively, it could represent a shift in allele frequencies of EBOV subpopulations within the survivor during the 20 days that passed between the date of sexual intercourse and potential transmission (March 7, 2015) and the date of semen collection (March 27, 2015). Nevertheless, the nearly identical EBOV genomes place the survivor and the patient in the same transmission chain, and case tracing confirms contact by vaginal intercourse.