Yellow fever virus is a mosquito-borne flavivirus present in tropical areas of Africa and South America. Although most infections are asymptomatic, clinical disease ranges from a mild febrile illness to a potentially fatal disease with hepatitis and haemorrhagic manifestations.¹ Yellow fever can be prevented by a live, attenuated viral vaccine (17D strain) that is recommended for people aged at least 9 months who are travelling to or living in areas at risk of yellow fever. Like all live vaccines, yellow fever vaccine is contraindicated for immunocompromised individuals, including recipients of solid organ transplants, because of an increased risk of vaccine-associated adverse events.²

In October, 2021, four recipients of solid organs from a common, deceased donor in the USA each developed a febrile illness followed by neurological symptoms 2–6 weeks after transplantation. We describe the results of an investigation into the cause of these neurological symptoms.

The investigation was conducted from Nov 7, 2021, to Feb 24, 2022, and included six transplant recipients (four who received solid organs [two kidney, one heart, one liver] and two who received corneas), one organ donor, and a blood traceback investigation.

The recipient of the right kidney was a woman in her 60s with a history of type 1 diabetes, hypertension, and a previous kidney and pancreas transplant. She had no history of travel in the month before transplantation or of yellow fever vaccination. For the current kidney transplant she received induction immunosuppression with high-dose steroids and basiliximab, followed by a maintenance regimen of tacrolimus, mycophenolate mofetil, and prednisone. 4 weeks after transplantation she was admitted to hospital with fever, fatigue, and malaise (figure 1). Within 6 days of hospitalisation, the patient’s cognitive status rapidly deteriorated and she developed left hemiparesis and unresponsiveness requiring intubation for airway protection. Multiple brain MRI scans showed chronic microvascular changes. Loss of brainstem reflexes was noted 10 days post-onset. A lumbar puncture showed 2 white blood cells per μL, 28 red blood cells per μL, normal glucose concentrations, and increased protein concentrations (table 1). Testing for multiple infectious causes of meningoencephalitis was negative (appendix pp 4–5). The patient’s clinical status remained unchanged despite treatment with high-dose steroids, plasmapheresis, and intravenous immunoglobulin G. After being treated in hospital for approximately 1 month, she was transferred to long-term acute care, where she remained ventilator-dependent with no neurological improvement and died 7 months after transplantation.

Clinical metagenomic next-generation sequencing of CSF collected 11 days post-onset was negative for pathogenic organisms, except for the detection of yellow fever virus at sub-threshold reporting levels (one read). The read was 100% identical across 106 base pairs to 17D yellow fever vaccine strain (figure 2). Deeper metagenomic sequencing did not identify any additional reads from yellow fever virus or other pathogens.

The CSF contained detectable yellow fever virus-specific IgM antibodies but not neutralising antibodies, and viral RNA was not detected by RT-PCR. Serum collected 6 days before transplantation had no yellow fever virus-specific antibodies, whereas serum collected 14 weeks after transplantation had yellow fever virus-specific IgM and neutralising antibodies, indicative of seroconversion (table 2). The host gene expression data generated from the metagenomic next-generation sequencing assay of the CSF aligned more closely with CSF transcriptional profiles from patients with viral meningoencephalitis than from patients with autoimmune or non-infectious encephalitis (figure 3).

The recipient of the heart was a man in his 50s with a history of hypertension, type 2 diabetes, ischaemic cardiomyopathy, and congestive heart failure. He had no history of travel in the month before transplantation and his vaccination history was unknown. The patient received induction immunosuppression with high-dose steroids and basiliximab, followed by a maintenance regimen of tacrolimus, mycophenolate mofetil, and prednisone. 17 days after transplantation he was admitted to hospital with fever, malaise, and lethargy, followed by a progressive decline in cognitive status (figure 1). A lumbar puncture 12 days post-onset showed mild lymphocytic pleocytosis and increased protein concentrations (table 1). Brain MRI showed chronic volume loss. Tacrolimus was switched to cyclosporine and then to sirolimus, owing to concerns of possible calcineurin inhibitor-related encephalopathy. However, the patient progressed to unresponsiveness with no movement of his extremities. A trial of high-dose steroids was initiated, which was followed by a transient increase in responsiveness, tracking, and the following of some commands. Subsequent treatment with plasmapheresis and intravenous immunoglobulin G did not result in further improvement. The patient’s family opted to withdraw support, and he died 56 days after transplantation.

Testing of CSF specimens was negative for pathogenic organisms, except for the detection of human polyomavirus 2 (JC virus) by PCR (appendix pp 5–6). CSF collected 19 days post-onset had detectable yellow fever virus-specific IgM antibodies but no detectable neutralising antibodies, and no viral RNA was detected by RT-PCR or metagenomic next-generation sequencing (table 2). The host gene expression data generated from the metagenomic next-generation sequencing assay aligned more closely with profiles from patients with viral meningoencephalitis than from patients with autoimmune or non-infectious encephalitis (figure 3).

No yellow fever virus-specific or flavivirus-specific RNA was detected by RT-PCR in FFPE heart biopsy tissue after transplantation (table 2). FFPE brain tissue obtained at autopsy showed encephalitis with multiple foci of perivascular mononuclear inflammation, mild gliosis, and patchy regions of neuronal necrosis (figure 4). Immunohistochemistry results were negative for yellow fever virus; however, viral RNA was detected by RT-PCR in blocks from the dentate, hippocampus, pons, and basal ganglia. Sequence analysis showed shared identities with 17D yellow fever vaccine strain (table 2). An in-situ hybridisation assay found that yellow fever virus RNA was localised in encephalitic lesions with punctate assay labelling, suggesting the presence of viral RNA in degenerating neuronal processes (figure 4). RNA extracted from the brain blocks had 15 sequences aligning with 17D yellow fever vaccine strain identified by metagenomic next-generation sequencing (table 2). In total, there was 8% coverage of the yellow fever virus genome representing several different genes, although the depth of coverage was low and was often a single read. Compared with the canonical 17D yellow fever vaccine strain genome, one amino acid mutation of an arginine to a glycine was identified in a single read in the envelope protein at position 52 (appendix p 10).

The recipient of the liver was a man in his 40s with a history of autoimmune hepatitis and cirrhosis. He had no history of travel in the month before transplantation or of yellow fever vaccination. After induction with basiliximab, the patient was on a maintenance immunosuppressive regimen of tacrolimus and mycophenolate mofetil. 15 days after transplantation, he developed fever, nausea, vomiting, abdominal pain, and malaise (figure 1). A CT scan of the patient’s chest, abdomen, and pelvis showed moderate pelvic ascites. 9 days post-onset, the patient had persistent fevers and developed headache, diplopia, photophobia, multidirectional nystagmus, dysmetria, and nuchal pain. Analysis of the CSF showed lymphocytic pleocytosis and increased protein concentrations (table 1). Brain MRI showed subtle diffusion uptake and T2 hyperintensity in the cerebellum. Extensive testing for infectious diseases was negative except for Clostridioides difficile toxin in stool (appendix pp 6–7). The patient declined neurologically with myoclonus, posturing, and unresponsiveness, leading to intubation for airway protection. Because autoimmune encephalitis was suspected he was started on high-dose dexamethasone, which was followed by the rapid resolution of fever. He continued to exhibit hyper-reflexia, nystagmus, and myoclonus, but opened his eyes to stimuli. He subsequently underwent five rounds of plasmapheresis. After the second round he began following commands, opening his eyes spontaneously, and myoclonus and hyper-reflexia were resolved. He was extubated, and the findings of his neurological examinations returned to normal except for residual bilateral symmetrical weakness. The patient became ambulatory and was discharged to home at 31 days post-onset. 1 year after transplantation he was doing well.

FFPE liver tissue from a post-transplantation biopsy showed non-specific changes of steatosis and chronic portal inflammation and was negative for multiple pathogens, including yellow fever virus by immunohistochemistry and flavivirus by RT-PCR (table 2, appendix pp 6–7). Serum collected 13 days post-onset was negative for yellow fever virus RNA by RT-PCR and metagenomic next-generation sequencing, for yellow fever virus-specific IgM antibodies, and for yellow fever virus by viral culture (table 2). CSF collected 14 days post-onset was negative for yellow fever virus RNA by RT-PCR but positive for IgM antibodies; an insufficient quantity was available to test for neutralising antibodies. Serum collected 57 days post-onset contained yellow fever virus-specific IgM antibodies and neutralising antibodies; a serum specimen collected 1 day before transplantation had no detectable yellow fever virus-specific antibodies, indicating seroconversion (table 2).

The recipient of the left kidney was a woman in her 40s with a history of type 2 diabetes, hypertension, coronary artery disease, and peripheral vascular disease. She had no history of travel in the month before transplantation or of yellow fever vaccination. Her induction regimen consisted of high-dose steroids and basiliximab followed by a maintenance regimen of tacrolimus, mycophenolate mofetil, and prednisone. 40 days after transplantation, she was admitted to hospital with fever, nausea, vomiting, and diarrhoea (figure 1). She was noted to have tremors and unsteady gait. CSF analysis showed 7 white blood cells per μL and normal protein concentration (table 1). Brain MRI showed mild dural thickening and enhancement. Extensive testing for infectious disease causes was negative (appendix p 8). The patient’s dose of mycophenolate mofetil was discontinued and she showed clinical improvement; she did not receive any treatment with intravenous immunoglobulin G. 1 year after transplantation she was doing well.

Metagenomic next-generation sequencing of CSF collected 4 days post-onset was negative for yellow fever virus RNA (table 2); however, the host gene-expression data generated from the CSF metagenomic next-generation sequencing assay aligned with the transcriptional profiles of the other two organ recipients and with the viral meningoencephalitis cohort (figure 3). Serum collected 31 days post-onset contained yellow fever virus-specific IgM antibodies and neutralising antibodies. Serum collected 8 months before transplantation had no yellow fever virus-specific antibodies, indicating seroconversion (table 2).

The recipient of the right cornea was a man in his 20s with a history of migraines who underwent penetrating keratoplasty of his cornea following ocular trauma. He was receiving topical but no systemic immunosuppressive medications. He had no history of travel in the month before vaccination or of yellow fever vaccination. 40 days after transplantation he developed intermittent headaches, nausea, and tongue tingling. He had no fever or other neurological symptoms. Testing of the CSF for multiple causes was negative (table 1, appendix pp 8–9). His headaches resolved, but because of the suspicion of infection in recipients of organs from the same donor and the presence of neurological symptoms in the patient, the cornea was removed and replaced. 10 months after transplantation the patient was doing well.

Testing of CSF, corneal explant tissue, and aqueous humour from the anterior chamber was negative for yellow fever virus RNA by metagenomic next-generation sequencing, and the host gene-expression data aligned more closely with that from patients with autoimmune or non-infectious diseases than with the CSF transcriptional profiles of the three organ recipients or the viral meningoencephalitis cohort (figure 3). FFPE corneal explant tissue was negative for yellow fever virus by immunohistochemistry and for flavivirus by RT-PCR. A follow-up serum specimen obtained 76 days after transplantation was negative for yellow fever virus-specific IgM antibodies (table 2).

The recipient of the left cornea was a woman in her 60s who had a partial thickness cornea transplant. The patient had no reported symptoms and was doing well 1 year after transplantation; no post-transplantation testing for yellow fever virus was conducted.

The organ donor was a man in his 50s with no clinically significant medical history who had a traumatic brain injury. The patient had no history of travel in the several months before transplantation or of yellow fever vaccination. During his 9-day treatment in hospital before organ donation, the patient received six blood products—five packed red blood cell units and one unit of platelets—from six donors. He was declared brain dead 5 days after admission. No signs or symptoms of neurological or other infection were present at the time of death.

FFPE tissue from needle biopsies of the liver and kidneys collected at procurement showed no histopathological findings suggestive of infection and were negative for multiple pathogens, including yellow fever virus (table 2). Pre-transplantation serum collected within 1 day of organ procurement and after the patient received all blood-product transfusions had no detectable yellow fever virus RNA or antibodies.

Five of six blood donors were reached and interviewed retrospectively. No donors reported any symptoms, international travel, or exposure to mosquitos within 30 days of donation. However, one donor reported receiving a primary yellow fever vaccine dose (YF-VAX; Sanofi Pasteur, Swiftwater, PA, USA) 6 days before donating blood (figure 1). During pre-donation screening, this blood donor did not report receiving any live vaccines within 8 weeks of donation. The donor provided a split apheresis red blood cell donation resulting in two red blood cell products. One red blood cell unit was transfused into the organ donor 24 days after the blood donation and 3 days before organ procurement (figure 1). The second unit was transfused into a different patient 26 days after the donation; no adverse events attributed to the vaccine were identified on review of this patient’s medical record. This second recipient died 10 weeks after the blood transfusion from complications related to cancer; no specimens from this patient were available for testing.

The US Vaccine Adverse Event Reporting System and the vaccine manufacturer were queried about any safety concerns associated with the specific vaccine lot administered to the blood donor; no increase in reported adverse events was noted.

Transmission of 17D yellow fever vaccine strain by blood transfusion to an organ donor and subsequent transmission through solid organ transplantation led to neurological disease in four organ recipients. Two recipients recovered and two died after neurological deterioration.

Yellow fever vaccine was originally developed in the 1930s through serial passage of Asibi wild-type yellow fever virus.¹ All currently available vaccines are live, attenuated vaccines using 17D viral substrains. Serious adverse events from the vaccine are rare but include acute anaphylactic reactions, yellow fever vaccine-associated viscerotropic disease (which resembles natural infection), and yellow fever vaccine-associated neurological disease.² As with all live vaccines, yellow fever vaccine is contraindicated in people with altered immune function from underlying conditions, such as thymus disorders, or people who are on immunosuppressive or immunomodulatory therapies, including organ transplant recipients.^2,14

Yellow fever vaccine-associated neurological disease is caused by either direct neuroinvasion of the virus into the CNS (neurotropic disease) or an autoimmune-mediated process triggered by the virus. The most common presentation of neurotropic disease is meningoencephalitis. Autoimmune-mediated neurological disease typically presents as Guillain-Barré syndrome, acute disseminated encephalomyelitis, or transverse myelitis; however, cases of autoimmune-mediated encephalitis have been reported.¹⁵ Symptom onset typically occurs 2 weeks after vaccination, but can range from 2 days to 50 days.¹⁶ Initial symptoms can include fever and headache followed by more specific signs and symptoms related to the clinical syndrome. Diagnosis is based on a combination of clinical, imaging, and laboratory findings and the absence of an alternative diagnosis.² Patients with meningoencephalitis typically have a good outcome with complete recovery, whereas older patients and those with other autoimmune manifestations can have more complications. Deaths with the condition are rare, with only three cases reported in the literature.^17–19

Yellow fever vaccine-associated neurological disease among the transplant recipients in this cluster was notably severe and led to deaths. Because yellow fever vaccination is contraindicated in the recipients of solid organ transplants, data on adverse events of the vaccine are scarce for this group. Although no serious adverse events have been described in organ recipients who inadvertently received the vaccine, most patients had received transplants several years previously, and the numbers of cases are too small to assess safety.^20–22

All organ recipients in this cluster were confirmed to have recent yellow fever virus infection. Neuroinvasive disease was confirmed in the recipient of the heart by molecular testing of brain tissue. The recipient of the right kidney had initial molecular evidence of yellow fever virus in the CSF by metagenomic next-generation sequencing. The recipients of both kidneys and the liver had evidence of seroconversion in peripheral blood samples. However, the recipients of the heart, liver, and right kidney had yellow fever virus-specific IgM antibodies in the CSF but no neutralising antibodies in the available specimens. Immunosuppression could explain the absence of or delay in neutralising antibody responses in the CSF. Another possible explanation is an autoimmune-mediated process, which was originally suspected in the recipient of the liver, who had autoimmune hepatitis; underlying autoimmune disease can be triggered by vaccines through activation of cellular and immune responses.¹⁵ However, host gene-expression signatures in the CSF of all three organ recipients in whom this was evaluated suggest a predominantly neuroinvasive pathway. Although using human gene-expression data for diagnosing neurological infections is still in its infancy, these results show the potential for host response-based profiling to complement traditional microbiological methods in diagnosing infectious diseases.^23,24

The pathogenesis leading to neurological rather than viscerotropic disease in all organ recipients in the present cluster is unclear. Early studies suggested that a low dose of vaccine might cause a delayed immune response and increased risk of encephalitis, although this has not been supported by subsequent data.¹ Reversion of the vaccine strain to neurovirulence, either by mutation in a variant virion in the vaccine lot (unlikely given the absence of safety signals identified) or in the blood or organ donor, was also considered. Such reversion was previously seen in a fatal case of yellow fever vaccine-associated encephalitis in which sequencing of virus from the brain revealed a mutation at position 303 of the envelope protein that is associated with increased neurovirulence in mice and monkeys.¹⁷ Sequencing of viral RNA from the brain tissue of the heart recipient identified one amino acid substitution at position 52 of the envelope protein, in which arginine reverted to glycine as in the Asibi wild-type strain. Although this mutation is not known to be associated with neurovirulence, it could cause a confirmational change that affects viral entry into host cells.¹

Transmission of 17D yellow fever vaccine strain through the transfusion of blood products has previously been reported with no adverse events identified in five recipients.²⁵ In immunocompetent individuals who have not previously been exposed to yellow fever wild-type or vaccine virus, peak viraemia occurs 3–5 days after vaccination and can last for up to 10 days after vaccination, by which time most people develop neutralising antibodies.¹ The blood donor in this investigation received the vaccine 6 days before donating blood, so a low-level viraemia could have been present at the time of donation. In this case, the standard pre-donation screening questionnaire—which included a question about recent vaccinations—was administered but failed to elicit vaccination history to prevent donation, showing the limitations of obtaining accurate information through the administration of donor history questionnaires. A previous cognitive evaluation of the uniform donor history questionnaire used in the USA revealed that donors often answer questions through the perspective of “is my blood safe?” rather than providing objective answers to questions.²⁶ As a result, individuals who would otherwise be deferred from donation might continue to donate blood, resulting in risk to recipients. Risk-mitigation strategies, such as pathogen reduction technology for apheresis platelets and plasma, are increasingly adopted by blood collection organisations and could be available for red blood cell products in the future, should such products be approved for use.^27,28

This investigation was limited by the lack of availability of crucial specimens. We were not able to test for yellow fever virus in the implicated donated blood because blood products were no longer available at the time of the investigation. Yellow fever virus was not detected in the organ donor, possibly related to dilutional effects of resuscitation measures, low-level viraemia below the detection threshold, or viraemia resolution and viral sequestration in the organs during procurement.²⁹ Low volumes of tissue from organ biopsies might have prevented the detection of yellow fever virus. These limitations were overcome for some recipients, in whom seroconversion was revealed by the testing of stored serum collected before transplantation and post-transplant samples. However, this investigation also highlights the need to ensure sample retention—both before transplantation and from the acute illness—to facilitate confirmatory testing.

The transmission of 17D yellow fever vaccine virus through solid organ transplantation illustrates the value of clinicians recognising the possibility of donor-derived infection and collaborating with public health authorities to complete a comprehensive evaluation. This finding also substantiates the clinical use of metagenomic next-generation sequencing for the broad-based detection of pathogens if a transplant-transmitted infection is suspected and traditional diagnostic testing has not identified a cause.²³ This cluster of cases occurred because a blood donation was made soon after receipt of a live virus vaccine, and the associated blood product was transfused into an organ donor. Although blood transfusion continues to become safer over time, transfusion-transmitted infection of an organ donor can rarely result in transplant-associated infections.^30,31 Continued efforts at haemovigilance, adoption of risk-mitigation strategies, and implementation of new safety interventions should reduce the risk of transfusion-transmitted infections.³² Healthcare workers providing yellow fever vaccination should inform patients of the need to defer blood donation for at least 2 weeks after vaccination.