Study Design and Settings

We conducted a retrospective and multicenter study in French Guiana, which covers an area of 83,846 km² and neighbors Brazil and Suriname. The private healthcare sector in French Guiana is limited; healthcare services are provided mainly by 3 hospital centers. CHC and Western Guianese Hospital Center (WGHC) are the 2 major healthcare centers (Figure 1). A total of 16 Delocalized Centers for Prevention and Care (DCPCs), which are a hospital department of CHC (i.e., sharing the same medical tools and software), oversee the primary care for persons in the Amazon areas of the territory. All samples collected in the DCPCs were sent to CHC, except samples from the DCPCs of Grand Santi, Apatou, Javouhey, and Awala Yalimapo, which were sent to WGHC by agreement.

Sociodemographic and Clinical Data Collection

We extracted all cases with a bacterial isolate belonging to the CdSC, laboratory diagnosed in CHC or WGHC during January 1, 2016–December 31, 2021, using HEXALIS results-reporting software (AGFA Healthcare). We arbitrarily considered positive samples for the same patient as 2 distinct episodes when they were collected >6 months apart and after clinical cure of the first episode or negative bacteriological control after >1 month of well-managed treatment.

We extracted patient sociodemographic and clinical data using the medical monitoring software (Cora Systems) for CHC and WGHC (including DCPC). We excluded patients without any clinical information found. Patients who objected to the use of their medical data in this study were secondarily excluded. We treated patient data that were not available in the medical records as unavailable data. For data describing patient progress, we considered patients who did not attend scheduled consultations as lost to follow-up (and not as unavailable data).

Bacterial Isolation, Identification, and Toxigenicity and Antimicrobial Susceptibility Testing

We identified and cultivated bacteria at CHC and WGHC on Columbia horse blood agar plate (bioMérieux) for 24 hours at 35°C–37°C. We performed bacterial identification by using Bruker matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. We routinely sent all CdSC isolates to the NRC at the Institut Pasteur of Paris for further analysis.

At NRC, isolates were grown and purified on Tinsdale agar and characterized for pyrazinamidase, urease, nitrate reductase, use of maltose, and glycogen fermentation (16). We used a 4-plex real-time PCR to detect the diphtheria toxin gene and confirm species (17). We assessed the phenotypic production of the toxin using a modified Elek test (18). We performed antimicrobial susceptibility testing by disc diffusion (Bio-Rad Laboratories) and used E-test for MIC determination if deemed resistant on the basis of disk diffusion (Appendix).

Genome Sequencing and Phylogenetic Analyses

We performed genomic analyses at NRC using Illumina technology and de novo assembly as previously described (16,19). We conducted the search for resistance genes and tox gene integrity using diphtOscan (20). We defined multilocus sequence type (MLST) and core-genome MLST (cgMLST) genotypes using the Institut Pasteur CdSC database (https://bigsdb.pasteur.fr/diphtheria). For C. diphtheriae isolates, we defined the sublineages (500 mismatches) and genetic clusters (25 mismatch threshold) groupings as previously proposed (19) (Appendix).

Statistical Analysis

We performed statistical analyses using Microsoft Excel and RStudio version 02.3 software. We expressed qualitative variables as numbers and percentages and distribution of quantitative variables as means and SDs. We calculated estimates of annual incidence per 100,000 inhabitants on the basis of demographic data published annually by the French National Institute of Statistics and Economic Studies during 2016–2021 (21−26) (Appendix).

Ethics Approval

In conformity with French legislation, this retrospective study not involving human persons adheres to the Reference Methodology MR-004, with CHC’s compliance commitment dated December 21, 2021. A privacy impact analysis was conducted; a study summary is available on the Health-Data-Hub (no. F20220825152116). Its legal basis is a public interest mission. Data were sourced from routine care patient medical files and anonymized. Regulatory steps were taken to inform patients and allow refusal (Appendix). This approach was conducted in compliance with regulations in Europe (https://www.cnil.fr/fr/reglement-europeen-protection-donnees).

Included Bacterial Isolates and Associated Clinical Cases

For the study period, we identified 64 cases of bacterial isolates of CdSC isolated from 60 patients. No patients objected to the use of their data for the study, and none were excluded (Figure 2). We found 61 cases of C. diphtheriae, corresponding to 58 cutaneous forms and 3 noncutaneous forms (1 superinfection of chronic nasal mucosal disease, 1 respiratory infection, and 1 endocarditis), and 3 cases of C. ulcerans, all of which were cutaneous forms. Of the 60 patients, 2 had 2 positive C. diphtheriae samples each, corresponding to unrelated cases of cutaneous infection, and 1 patient had 3 positive C. ulcerans samples, corresponding to 3 different cases.

A total of 5 C. diphtheriae cases were tox-positive by quantitative PCR (qPCR) (8%; 5/64). However, they did not produce the diphtheria toxin; they were all negative by Elek’s test. Those 5 cases thus correspond to nontoxigenic tox gene–bearing isolates.

Evolution of Estimated Incidence during Study Period

The number of annual cases has risen steadily. Only 2 cases were diagnosed in 2016 and 2017, but in 2022, the number of annual cases reached 22. The estimated incidence has risen from 0.7 cases/100,000 inhabitants/year in 2016 to 7.7 cases/100,000 inhabitants/year in 2021 (Figure 3).

Epidemiology, Sociodemographics, and Medical History of Patients

In total, 80% (51/64) of the cases in our study were diagnosed in a DCPC by initial bacteriological sampling, whereas the others were mainly geographically dispersed (Figure 1). Mean patient age was 30.4 (+23.7) years (interquartile range 10.5–49.5) the male-to-female ratio was 1.7:1 (38/22). Of the 55 patients for whom information was available, 19 (35%) did not have social security coverage at the time of care, and 3 (5%) were homeless.

Of 58 patients with available data, 6 (10%) had a history of relative immunosuppression (diabetes, n = 3; alcoholic cirrhosis, n = 2; panhypopituitarism, n = 1). In addition, 10 (17%) were alcoholics, smokers, or polydrug addicts; 3 (5%) had a psychiatric pathology; and 8 (14%) had a chronic skin pathology that predisposed them to wounds or superinfections. Of 44 patients with available vaccination data, 29 (66%) were up-to-date with diphtheria vaccination at the time of infection (in accordance with French Public Health requirements for diphtheria) (27).

Clinical Description of Cases Caused by C. diphtheriae

Among the 61 cases caused by C. diphtheriae, 58 were cutaneous infections and 3 noncutaneous infections. Of the 58 cutaneous infections, the infection occurred on an existing wound in 23 (40%) persons: 6 occurred in a chronic skin lesion (lasting >6 weeks) and 17 occurred in an acute lesion (lasting <6 weeks). The infections mainly involved the lower limbs (62%, 36/58), followed by the head (12%, 7/58), the abdomen (7%, 4/58) and the upper limbs (2%, 1/58); 17% of cases extended to >2 areas of the body (10/58). In 57% (33/58) of cases, several lesions were present; in 25 cases, fibrinous involvement was described (56%, 25/45; data were not available for 13 cases). One patient, in whom 2 cases of cutaneous infection were present, had a background of immunosuppression; those 2 cases occurred on a chronic cutaneous lesion.

Data regarding clinical management were available for 52 cases. For 6 (12%) cases, treatment consisted only of simple dressing care. The 46 (88%) other cases were also treated by antibiotics; 40 were treated with amoxicillin (16 for 14 days, 16 for <10 days; data were not available for 8 cases), 2 were treated with azithromycin for 3 days, and 4 were treated with ciprofloxacin, pristinamycin, or ceftriaxone. One patient with multiple abscesses underwent surgery. The 5 patients with tox-positive isolates had cutaneous infection without toxic manifestations. No patients received diphtheria antitoxin while results of the qPCR were pending because they had no signs of severity attributable to toxin expression. For the 58 patients with C. diphtheria cutaneous infections, the outcome was favorable for 22 persons, 4 were lost to follow-up, and information was missing for 25. The clinical outcome was described as unfavorable for 7 patients (i.e., 1 patient died, 2 had a recurrence after 1 month, ulcerations were described as persistent despite well-managed treatment for 2 others, and details were not available for 2 more). For 18 cases (44%, 18/41 with available data among the 58 C. diphtheriae cutaneous cases), an investigation was conducted into the case.

For the 3 noncutaneous infections, the first patient (68 years of age) had a lower respiratory infection; Staphylococcus aureus, Pseudomonas aeruginosa, and Corynebacterium diphtheriae were also found in the bronchial aspirate. The patient died of septic shock caused by Klebsiella pneumoniae. The second patient, a 2-year-old child with a polymalformative syndrome, had a diagnosis of endocarditis to C. diphtheriae (confirmed by blood sample and secondarily confirmed by cardiac biopsy). The third patient (72 years of age) experienced a nasal mucosal form (superinfection of a chronic injury confirmed by mucus sampling). For those last 2 patients, details on the nature of the treatment were not available, but the clinical outcome was favorable. The diphtheria vaccination status of those patients was unknown.

3 Cutaneous Cases Caused by C. ulcerans in 1 Patient

One patient (74 years of age), who did not live with pets, was up-to-date on diphtheria vaccination, and had an underlying chronic skin wound (lymphatic filariasis sequelae), experienced 3 unrelated cases at 1-year intervals; eradication was monitored by samples collected 2 months after acute and treated cases. He had several lesions with fibrinous involvement located in the lower limbs (2 cases) and in upper limbs (1 case). Treatment consisted of amoxicillin for the first infection (unknown duration) and trimethoprim/sulfamethoxazole (for 10 days) for the second one; treatment for the last infection was not specified. The clinical course was described as good for all infections. Investigation into the case was not documented. The 3 C. ulcerans isolates belonged to the same genotype (sequence type [ST] 719) and were almost identical at the genomic sequence level (Appendix Figure).

Description of Coinfections

Of the 64 samples included, 61 (95%) were associated with bacterial co-infections. Among those 61 co-infections, 44 (69%) isolates were associated with a Streptococcus sp., 45 (70%) with S. aureus, and 31 (48%) with both of those 2 species; 15 (23%) were co-infected with another bacterium. The 3 non–co-infected cases corresponded to a blood culture sample and 2 cutaneous forms.

Genomic Epidemiology of Isolates

We studied the genomic diversity of 63 sequenced isolates (1 C. diphtheriae isolate could not be sequenced). The phylogenetic structure of the C. diphtheriae isolates from French Guiana formed 2 main lineages (Figure 4); 26 isolates corresponded to biovar Gravis (and its spuA marker gene) and 32 corresponded to biovar Mitis (16). Two isolates were biovar Belfanti (nitrate negative); for 1 of those, a disruptive mutation on the narH gene involved in nitrate reduction was found.

For C. diphtheriae isolates, sublineage-level classification showed 16 distinct sublineages (SL; defined using the 500 cgMLST mismatch threshold) (Figure 4). The most frequent SL was SL536 (6.6%, 11/63; all ST536). The other SLs had on average 3.3 isolates (SD 2.1). On the finer classification level, 26 genomic clusters were identified (defined using the 25 cgMLST mismatch threshold). One genomic cluster, GC295, was collected 9 times and belongs to the predominant SL536 (Figure 4). The 5 tox-positive isolates (but nontoxigenic because of a stop codon) belonged to the same sublineage and genomic cluster (ST67, all biovar Mitis).

When reanalyzing 1,350 global C. diphtheriae isolates (20), we found only 3 sublineages of this study were also observed outside of French Guiana: in metropolitan France, Brazil, or Malaysia. At the genomic cluster level, only 1 French Guiana genotype (GC341) was also observed in metropolitan France, consistent with this patient having recently traveled there. Most isolates (41/63) were susceptible to all antimicrobial drugs tested (Appendix).