The prevalence of acute septic arthritis in Western Europe is ≈4–10 cases/100,000 inhabitants (1). We report a case of posttraumatic knee septic arthritis in an immunocompetent patient in France that was caused by Erwinia billingiae, a gram-negative environmental bacterium of the family Enterobacteriaceae. We also review the characteristics of Erwinia species and infections.

On April 9, 2017, a 65-year-old man with an unremarkable medical history was admitted to an emergency unit in Nice, southern France, for painful right knee swelling that occurred a few hours after a Phoenix palm tree needle pierced the area. The foreign body was partly removed, and the wound was sutured. The patient was discharged without any knee pain and given a prescription for amoxicillin/clavulanic acid (1 g 3×/d for 6 d).

On April 22, the patient was admitted to the emergency unit of our hospital in Paris because of sudden right knee pain and fever. Synovial fluid collected by knee puncture the day of his admission to the orthopedic unit (April 23) contained 118 × 10⁹ leukocytes/L, consisting of 64% polymorphonuclear cells, 33% lymphocytes, and 3% other leukocytes; no microorganism could be identified after Gram staining and cultures. A second knee puncture was performed 3 days after admission, and gram-negative rods grew within 2 days solely within the anaerobic blood culture vial (BacT/ALERT SN; bioMérieux, https://www.biomerieux.com). Subcultures of the blood culture vial were positive after 24 hours of incubation at 37°C on blood agar (Trypticase Soy agar + 5% horse blood and Mueller Hinton 2 agar + 5% sheep blood; bioMérieux) and Drigalski agar (BD, https://www.bd.com) under aerobic conditions and chocolate agar (BD) under microaerobic conditions.

Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (Bruker Daltonik, https://www.bruker.com) was performed on colonies and failed to correctly identify the species. Therefore, we performed species identification by 16S rRNA amplification and sequencing with primers RNA-S (16S, 5′-AGAGTTTGATCCTGGYTCAG-3′) and RNA-AS (16AS, 5′-CTTTACGCCCARTAAWTCCG-3′) at a hybridization temperature of 52°C. We amplified a 521-bp sequence that matched the E. billingiae genome of 2 isolates with 99.4% similarity (GenBank accession nos. JQ929658 and JN175337). Other closely related species displayed lower similarities: Pantoea rwandensis (99.0%), Erwinia persicina (98.9%), Pantoea coffeiphila (98.7%), Erwinia tasmaniensis (98.5%), and Erwinia aphidicola (98.3%). Following guidelines of the Antibiogram Committee of the French Society for Microbiology (https://www.sfm-microbiologie.org/2019/01/07/casfm-eucast-2019), we tested the E. billingiae isolate with the antimicrobial drugs recommended for Enterobacteriaceae; the isolate was susceptible to all these drugs, including ampicillin.

Because of the lack of clinical improvement, the joint was washed on day 6 after admission. After this intervention, an empiric antimicrobial drug treatment was started with amoxicillin/clavulanic acid (2 g 3×/d intravenously). Once results of drug susceptibility testing became available (i.e., 10 days after admission), his treatment was switched to cefotaxime (2 g 3×/d intravenously) and ciprofloxacin (500 mg 2×/d orally for 8 d), followed by ciprofloxacin (500 mg 2×/d alone for 38 additional days). Total duration of treatment was 45 days. The clinical evolution of this patient was favorable; he fully recovered and had no relapses up to 1 year after treatment completion.

In the past, some members of the Erwinia genus were reassigned to the genera Enterobacter or Pantoea. Erwinia spp. are ubiquitous in the environment, especially in water ecosystems and soils. Plant-associated Erwinia species comprise epiphytic nonpathogenic (i.e., E. billingiae and E. tasmaniensis) and pathogenic (i.e., E. amylovora and E. pyrifoliae) species. The MALDI-TOF mass spectrometry system failed to identify the bacterium, even though E. billingiae is contained in the database for either method used (direct deposit or on-plate formic acid treatment). Future expansion of the database with more spectra will likely improve the performance of the MALDI-TOF mass spectrometry system for E. billingiae identification. Indeed, the database contains fewer spectra of E. billingiae (n = 4) than those of frequently encountered species in medical microbiological laboratories, such as Escherichia coli (n = 14) and Staphylococcus aureus (n = 10).

To further investigate Erwinia infections in humans, we reviewed reports available in PubMed published during 1967–2017 written in English by using the keywords “Erwinia” and “infection” (Table). Among the 17 cases reported, the sites of infection were diverse, and most (53%, 9/17) cases occurred after a direct inoculation during an injury with a plant (Table). We found no reports of osteoarticular infections with Erwinia; the only other E. billingiae case reported was a dermohypodermitis (Table). In that case, as in the case we report here, an injury with a plant was reported.

This case report illustrates the importance of the methods used for bacterial identification to correctly diagnose such infections. Biochemical methods (2–8) and MALDI-TOF mass spectrometry (as done in our investigation) could result in misidentification. This report highlights the usefulness of analyzing MALDI-TOF mass spectrometry scores before assigning a species identity and sequencing the 16S RNA gene for bacteria not identifiable by conventional methods.

Top