1. Background and Scenario
This report documents the analysis of a fictional multi-hospital outbreak of Carbapenem-Resistant Acinetobacter baumannii (CRAB) designed as part of the EURGen-RefLabCap Multidisciplinary WGS Training Workshop, conducted virtually in January 2024. The exercise was coordinated by Statens Serum Institut / DTU and challenged National Reference Laboratories (NRLs) from across Europe to integrate epidemiological and whole-genome sequencing (WGS) approaches to investigate a complex, multi-hospital outbreak scenario.
Scenario Setting
A fictional National Reference Laboratory (NRL) in Country W receives clinical and epidemiological records for 71 Acinetobacter baumannii cases from six hospitals (Hospitals A–F) spanning January to December 2023. The NRL must determine whether an outbreak is occurring, how many clusters exist, which isolates are related, and how transmission may have spread between hospitals — potentially via patients with travel history to Country Z.
All patient names, hospital identifiers, countries, and isolate data in this exercise are entirely fictitious, created specifically for training purposes. No real patient data or real outbreak information is represented here.
Pathogen Profile: CRAB
Acinetobacter baumannii is a Gram-negative opportunistic pathogen classified by the WHO as a critical-priority organism for the development of new antibiotics. Carbapenem-resistant strains (CRAB) — particularly those carrying OXA-type carbapenemases (most commonly OXA-23) — pose a severe threat in healthcare settings due to their ability to cause difficult-to-treat infections and their propensity for nosocomial transmission. The Global Clone 2 (GC2) lineage, typed as ST2 by the Pasteur MLST scheme, is the most globally disseminated CRAB clone and is the dominant lineage in this exercise.
2. Bioinformatics Methods
Isolate Selection (Inject 1.1 / 1.2)
From the initial 71 clinical records, 36 priority isolates were selected for WGS based on the following metadata criteria:
- Hospital of admission and ward (ICU vs. general ward)
- Sample collection date and patient hospitalization/discharge dates
- Antibiogram pattern (especially carbapenem resistance profile)
- Carbapenemase genotype (OXA-23, NDM, or unknown)
- Patient travel history — particularly travel to Country Z
- Patient transfer history between hospitals
WGS Analysis Pipeline
Note on AB_43
Isolate AB_43 (Hospital B, Inject 3.1) had raw FASTQ reads available but no assembled FASTA, likely due to low coverage or failed assembly. It was excluded from the core-genome alignment and is therefore absent from the phylogenetic tree and SNP matrix. The remaining 35 isolates were used for phylogenetic reconstruction.
3. Inject 1 — Epidemiological Analysis
Introduction and Case Definition
Received: January 2024 · 71 Acinetobacter baumannii clinical recordsThe NRL is presented with 71 A. baumannii clinical records from six hospitals (A–F) spanning all of 2023. The initial task is to define a case, construct an epidemic curve, and select priority isolates for WGS.
Working case definition: A patient with a clinical A. baumannii isolate demonstrating carbapenem resistance (or confirmed OXA-23/NDM carriage), admitted to Hospital A–F during January–December 2023.
Epidemiological Data — EpiCurve Construction
71 clinical records with hospitalization dates, antibiograms, and travel historyExamination of the full case records reveals a temporal clustering of cases in Hospital A beginning in July 2023, with cases extending through the rest of the year. A subset of cases from Hospitals B–F appear later, some with documented patient transfers or travel to Country Z.
Key Metadata Prioritised for Isolate Selection
Workshop participants were asked to rank metadata factors. The most-selected indicators (Survey 1 results, N=20 participants) were:
Based on combined epidemiological assessment, 36 isolates (AB_23–AB_58) were prioritised for WGS, covering all six hospitals and including isolates from patients with Country Z travel history.
Epidemic Curve Interpretation
The epidemic curve for Hospital A shows a propagated (person-to-person) pattern emerging from July 2023, consistent with nosocomial transmission rather than a common-point source. The broader pattern across all six hospitals suggests inter-hospital spread possibly linked to patient transfers and importation events from Country Z.
4. Inject 2 — Hospital A WGS Analysis (18 Isolates)
Hospital A Sequencing Data Released
Isolates AB_23 – AB_40 · FASTQ + FASTAWhole-genome sequencing data for all 18 Hospital A isolates (AB_23–AB_40) are released. Participants perform MLST, AMR annotation, cgMLST, and SNP-based phylogenetic analysis.
MLST Results — Hospital A
| Isolate | MLST Pasteur | MLST Oxford | cgMLST | Carbapenemase | Status |
|---|---|---|---|---|---|
| AB_23, 24, 26, 27, 28, 29, 30, 31 | ST2 | ST208 | CT1451 | OXA-23 | Cluster A |
| AB_25 | ST1017 | ST1017 | — | OXA-66 | Sporadic |
| AB_32 | ST2 | ST195 | CT607 | OXA-23 | Inter-cluster |
| AB_33, 34, 35, 37, 38, 39, 40 | ST2 | ST195 / ST1816 | CT607 / CT2896 | OXA-23 | Cluster B |
| AB_36 | ST2 | ST195 | CT6 | NDM-2 | Sporadic |
Two Intra-Hospital Clusters Identified
AB_23, 24, 26, 27, 28, 29, 30, 31
- MLST: ST2 (Pasteur) / ST208 (Oxford)
- cgMLST: CT1451 (all identical)
- Carbapenemase: OXA-23
- Intra-cluster SNPs: 4–210 (closest pairs: AB_23/AB_29 = 4 SNPs; AB_26/AB_27 = 4 SNPs)
- Epi link: All from Hospital A with overlapping admission dates; direct patient-to-patient contact chains documented
AB_33, 34, 35, 37, 38, 39, 40 (±AB_32)
- MLST: ST2 (Pasteur) / ST195 or ST1816 (Oxford)
- cgMLST: CT607 / CT2896
- Carbapenemase: OXA-23
- Intra-cluster SNPs: ~1,100–3,600 (more diverse; possible indirect transmission)
- Epi link: Hospital A, overlapping admissions; patient AB_33 (PT11) linked to multiple cluster members
ST1017 / OXA-66 only
Genetically distant from both clusters (millions of SNPs across the full genome — completely different MLST lineage). Not part of the outbreak. Likely healthcare-associated acquisition with a different source.
ST2 / NDM-2 (no OXA-23)
Same species and MLST but carries NDM-2 rather than OXA-23. Phylogenetically distinct (~38,000 SNPs from Cluster A). Represents an independent CRAB case, not part of the OXA-23 outbreak.
Survey 2 Results — European NRL Consensus on Hospital A
18 of 19 responding NRLs identified 2 clusters in the Hospital A data. 100% of respondents identified signs of intra-hospital transmission. The most important metadata parameters selected were: hospitalization dates, MLST, cgMLST, carbapenemase genotype, and phylogenetic relation.
5. Inject 3 — Multi-Hospital Expansion
Hospital B & C Sequencing (10 Isolates)
Isolates AB_41 – AB_50 · Hospitals B and C · Including Country Z patientsSequencing data for 10 additional isolates from Hospitals B and C are released. Key isolates of interest include those from patients with documented travel to Country Z.
| Isolate | Hospital | MLST | Carbapenemase | Country | cgMLST | Finding |
|---|---|---|---|---|---|---|
| AB_41, AB_42 | Hospital B | ST2 | OXA-23 | Country Z | CT1414 | Outbreak — linked to Clusters A/B |
| AB_43 | Hospital B | — | — | — | — | No assembly (excluded from tree) |
| AB_44, AB_45 | Hospital B | ST103 | NDM-4 | Country Z | — | Sporadic — distinct lineage |
| AB_46, AB_47 | Hospital C | ST2 | OXA-23 | Country Z | CT1451 | Outbreak — Cluster A related |
| AB_48, AB_49 | Hospital C | ST2 | OXA-23 | Country Y | CT1451 | Outbreak — Cluster A related |
| AB_50 | Hospital C | ST2 | OXA-23 | Country X | CT607 | Cluster B–related / sporadic |
The phylogenetic tree integrating Hospitals B and C isolates confirms two critical findings:
- AB_46 and AB_47 (Hospital C, Country Z patients) cluster tightly with Hospital A Cluster A isolates — demonstrating inter-hospital spread of the ST2/OXA-23/CT1451 clone, plausibly via patient transfer or healthcare worker movement.
- AB_44 and AB_45 (Hospital B, Country Z patients, ST103/NDM-4) are phylogenetically completely distinct. These represent an independent importation event unrelated to the main outbreak. Their branch length in the ML tree is exceptionally long (~1.17 ML distance units), confirming they are genuine outliers.
Country Z — A Complex Signal
Multiple patient sub-groups from Country Z are present in the data. Survey responses found that participants identified: sporadic cases (16/19), intra-hospital transmission (9/19), and inter-hospital transmission (5/19) among Country Z-associated patients. The key distinction is that some Country Z patients carried the same ST2/OXA-23 outbreak clone (AB_41, AB_42, AB_46, AB_47), while others carried an entirely different NDM-4 lineage (AB_44, AB_45).
Hospital D, E & F Sequencing (8 Isolates)
Isolates AB_51 – AB_58 · Completing the full dataset| Isolate | Hospital | MLST | Carbapenemase | Finding |
|---|---|---|---|---|
| AB_51, 52, 53, 54, 55, 56 | Hospital D | ST2 | OXA-23 | Outbreak — Cluster A/B related |
| AB_57, AB_58 | Hospital E & F | ST2 | OXA-23 | Outbreak — Cluster A/B related |
The final inject completes the pan-hospital picture. Hospital D isolates (AB_51–AB_56) place within the Cluster A/B supercluster, confirming that the ST2/OXA-23 CRAB outbreak extended well beyond Hospital A. Isolates AB_57 and AB_58 (Hospitals E and F, respectively) also cluster within the outbreak clade, with bootstrap support of 100.
6. Full Genomic Results
6.1 MLST and Lineage Summary
| MLST Pasteur | Count | Carbapenemase | Interpretation |
|---|---|---|---|
| ST2 | 33 | OXA-23 (31), NDM-2 (2) | Main outbreak lineage (GC2); NDM-2 is sporadic |
| ST1017 | 1 | OXA-66 (intrinsic only) | Unrelated sporadic case (AB_25) |
| ST103 | 2 | NDM-4 | Unrelated Country Z importation (AB_44, AB_45) |
6.2 Phylogenetic Structure
The IQ-TREE2 maximum likelihood phylogeny (GTR+G, 1,000 ultrafast bootstrap replicates) resolves four major groupings:
- Main outbreak supercluster (bootstrap 100): Contains all 31 OXA-23-positive ST2 isolates across Hospitals A–F. Within this, two sub-clusters are supported:
- Cluster A: AB_23, 24, 26–31, 46–49, 51–58 — tight ST2/CT1451 clade, closest pairs differ by as few as 4 pairwise SNPs
- Cluster B: AB_32–35, 37–42 — broader ST2/CT607/CT2896 diversity, consistent with a longer or more indirect transmission chain
- AB_36 (NDM-2, ST2): Branches off within the ST2 tree but distinct from both clusters (~38,000 SNPs from Cluster A)
- AB_44 / AB_45 (NDM-4, ST103): Extremely long branch (ML distance ~1.17) — completely unrelated to outbreak
- AB_25 (ST1017): Outgroup — distinct lineage confirming this is a sporadic non-outbreak case
6.3 AMR Gene Profile
Abricate (CARD database) identified 49 unique resistance genes across 36 isolates (1,017 total hits). The major resistance gene categories are:
Core AMR Signature of Outbreak Isolates
All outbreak cluster isolates (Clusters A and B) share a common AMR backbone: OXA-23 (acquired), OXA-66 (intrinsic), ADC-30, adeABC, adeIJK, tet(B), and sulfonamide resistance genes — consistent with a pan-drug-resistant Hospital clone. AB_25 (ST1017) lacks OXA-23 and carries only the intrinsic A. baumannii resistome. AB_44/AB_45 (ST103/NDM-4) carry a completely different AMR gene set.
6.4 Reconstructed Transmission Timeline
7. Conclusions
Primary Conclusion
This exercise demonstrates a confirmed multi-hospital outbreak of ST2 (GC2) OXA-23-positive CRAB originating in Hospital A and spreading to all five additional hospitals. The outbreak involved at least two phylogenetically distinct sub-clusters (Cluster A and Cluster B) within the ST2/OXA-23 lineage. Inter-hospital transmission is strongly supported by phylogenomics and is partly linked to patient movement and travel from Country Z.
Outbreak clone confirmed
33/36 isolates belong to ST2 (Pasteur) with OXA-23 carbapenemase. All cluster within a single ML phylogenetic clade with bootstrap support of 100.
Multi-hospital spread
The ST2/OXA-23 clone was detected in all six hospitals. Inter-hospital links are supported both by genomic clustering and by patient transfer / travel history data.
Country Z — complex signal
Country Z-associated patients carried two distinct organisms: the outbreak clone (ST2/OXA-23, Cluster A) AND an unrelated NDM-4 strain (ST103). Careful genomic discrimination was essential to avoid misclassifying the NDM-4 cases as part of the outbreak.
Sporadic cases identified
AB_25 (ST1017), AB_36 (NDM-2), AB_44/45 (NDM-4/ST103), and AB_50 are not part of the outbreak. WGS correctly excluded these from the transmission cluster that epi data alone might have grouped together.
WGS adds critical resolution
MLST alone (all isolates = ST2) could not distinguish the two sub-clusters or the sporadic NDM-2 case. cgMLST and SNP-level analysis were required for definitive cluster assignment.
European NRL consensus
18/19 participating NRLs independently reached the same conclusion: 2 intra-hospital clusters in Hospital A, with 100% agreement on intra-hospital transmission — validating the robustness of WGS-based outbreak analysis.
Key Lessons from the Exercise
- Integrate epi and genomics early: Travel history, hospitalization dates, and patient transfers were essential for correct interpretation of WGS data — neither data type alone was sufficient.
- Do not conflate same-MLST with same-outbreak: ST2 includes both the main outbreak clone and the sporadic NDM-2 strain (AB_36). Carbapenemase genotype and SNP-level distances are needed to distinguish them.
- Long branches are informative: AB_25, AB_44, and AB_45 have extraordinarily long phylogenetic branches — an immediate visual signal of genetic distinctiveness requiring no statistical test.
- Bootstrap confidence matters: The main outbreak clade has 100% bootstrap support; internal nodes within clusters have lower (15–63%) support, reflecting genuine phylogenetic uncertainty where transmission chains are ambiguous.
- Reference selection in core-genome tools: Parsnp uses one assembly as a reference; the reference must be carefully chosen and should not be treated as a biologically meaningful root.
- cgMLST complements SNP analysis: CT types provide a reproducible, nomenclature-based shorthand that is consistent across laboratories, while SNP distances give higher-resolution discrimination within the same CT type.