Isothermal Microcalorimetry Improves the Time to Diagnosis of Fracture-related Infection Compared with Conventional Tissue Cultures

Academic Article


  • BackgroundA consensus definition recently was formulated for fracture-related infection, which centered on confirmatory criteria including conventional cultures that take time to finalize and have a 10% to 20% false-negative rate. During this time, patients are often on broad-spectrum antibiotics and may remain hospitalized until cultures are finalized to adjust antibiotic regimens.Questions/purposes(1) What is the diagnostic accuracy of isothermal microcalorimetry, and how does its accuracy compare with that of conventional cultures? (2) Does isothermal microcalorimetry decrease time to detection (or diagnosis) of fracture-related infection compared with conventional cultures? (3) Does isothermal microcalorimetry have a diagnostic accuracy or time advantage over conventional cultures in patients on chronic suppressive antibiotics?MethodsBetween July 2020 and August 2021, we treated 310 patients with concerns for infection after prior fracture repair surgery. Of those, we considered all patients older than 18 years of age with fixation hardware in place at the time of presentation as potentially eligible. All included patients returned to the operating room with cultures obtained and assessed by both isothermal microcalorimetry and conventional cultures, and all were diagnosed using the consensus criteria for fracture-related infection. Based on that, 81% (250 of 310) of patients were eligible; a further 51% (157 of 310) were excluded because of the following reasons: the capacity of the isothermal microcalorimetry instrument limited the throughput on that day (34% [106 of 310]), they had only swab cultures obtained in surgery (15% [46 of 310]), or they had less than 3 months follow-up after surgery for infectious concerns (2% [5 of 310]), leaving 30% (93 of 310) of the originally identified patients for analysis. We obtained two to five cultures from each patient during surgery, which were sent to our clinical microbiology laboratory for standard processing (conventional cultures). This included homogenization of each tissue sample individually and culturing for aerobic, anaerobic, acid-fast bacilli, and fungal culturing. The remaining homogenate from each sample was then taken to our orthopaedic research laboratory, resuspended in growth media, and analyzed by isothermal microcalorimetry for a minimum of 24 hours. Aerobic and anaerobic cultures were maintained for 5 days and 14 days, respectively. Overall, there were 93 patients (59 males), with a mean age of 43 ± 14 years and a mean BMI of 28 ± 8 kg/m2, and 305 tissue samples (mean 3 ± 1 samples per patient) were obtained and assessed by conventional culturing and isothermal microcalorimetry. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of isothermal microcalorimetry to diagnose fracture-related infection were compared with conventional cultures using a McNemar test based on the consensus definition of fracture-related infection. This consensus criteria is comprised of two levels of certainty for the diagnostic variables. The first is confirmatory criteria, where infection is considered definitely present and includes the presence of fistula/sinus tract/wound breakdown, purulent drainage or the presence of pus, presence of microorganisms in deep tissue specimens on histopathologic examination, presence of more than five neutrophils/high-powered field by histopathologic examination (only for chronic/late onset cases), and identification of phenotypically indistinguishable pathogens by conventional culture from at least two separate deep tissue/implant specimens. The second is suggestive criteria in which further investigation is required to achieve confirmatory status. Fracture-related infection was diagnosed for this study to minimize subjectivity based on the presence of at least one of the confirmatory criteria as documented by the managing surgeon. When suggestive criteria were present without confirmatory criteria, patients were considered negative for fracture-related infection and followed further in clinic after surgical exploration (n = 25 patients). All 25 patients deemed not to have fracture-related infection were considered infection-free at latest follow-up (range 3 to 12 months). The time to detection or diagnosis was recorded and compared via the Mann-Whitney U test.ResultsUsing the consensus criteria for fracture-related infection, there were no differences with the numbers available between isothermal microcalorimetry and conventional cultures in terms of sensitivity (87% [95% confidence interval 77% to 94%] versus 81% [95% CI 69% to 89%]), specificity (100% [95% CI 87% to 100%] versus 96% [95% CI 79% to 99%]), PPV (100% [95% CI 90% to 100%] versus 98% [95% CI 89% to 99%]), NPV (74% [95% CI 60% to 84%] versus 65% [95% CI 52% to 75%]), or accuracy (90% [95% CI 83% to 96%] versus 85% [95% CI 76% to 91%]; p = 0.13). The concordance by sample between conventional cultures and isothermal microcalorimetry was 85%. Isothermal microcalorimetry had a shorter median (range) time to detection or diagnosis compared with conventional cultures (2 hours [0.5 to 66] versus 51 hours [18 to 147], difference of medians 49 hours; p < 0.001). Additionally, 32 patients used antibiotics for a median (range) duration of 28 days (7 to 1095) before presentation. In these unique patients, there were no differences with the numbers available between isothermal microcalorimetry and conventional cultures in terms of sensitivity (89% [95% CI 71% to 98%] versus 74% [95% CI 53% to 88%]), specificity (100% [95% CI 48% to 100%] versus 83% [95% CI 36% to 99%]), PPV (100% [95% CI 85% to 100%] versus 95% [95% CI 77% to 99%]), NPV (63% [95% CI 37% to 83%] versus 42% [95% CI 26% to 60%]), or accuracy (91% [95% CI 75% to 98%] versus 78% [95% CI 57% to 89%]; p = 0.17). Isothermal microcalorimetry again had a shorter median (range) time to detection or diagnosis compared with conventional cultures (1.5 hours [0.5 to 48] versus 51.5 hours [18 to 125], difference of medians 50 hours; p < 0.001).ConclusionGiven that isothermal microcalorimetry considerably decreases the time to the diagnosis of a fracture-related infection without compromising the accuracy of the diagnosis, managing teams may eventually use isothermal microcalorimetry - pending developmental improvements and regulatory approval - to rapidly detect infection and begin antibiotic management while awaiting speciation and susceptibility testing to modify the antibiotic regimen. Given the unique thermograms generated, further studies are already underway focusing on speciation based on heat curves alone. Additionally, increased study sizes are necessary for both overall fracture-related infection diagnostic accuracy and test performance on patients using long-term antibiotics given the promising results with regard to time to detection for this groups as well.Level of EvidenceLevel II, diagnostic study.
  • Authors

    Digital Object Identifier (doi)

    Author List

  • Cichos KH; Spitler CA; Quade JH; Johnson JP; Johnson MD; Ghanem ES
  • Start Page

  • 1463
  • End Page

  • 1473
  • Volume

  • 480
  • Issue

  • 8