Abstract
Background
Septic arthritis diagnostic is an emergency which implies a treatment with antibiotics and hospitalization. The diagnosis is based on the cytobacteriological examination of the synovial fluid (SF), but direct bacteriological examination is insensitive, and the result of the culture is obtained only after several days. Therefore, there is still a need for a rapid, simple and reliable method for the positive diagnosis of septic arthritis. Such method must allow avoiding both unrecognized septic arthritis leading to major functional consequences, and overdiagnosis that will induce unnecessary expensive hospitalization and unjustified treatment. Mid-infrared (MIR) spectroscopy, that gives a metabolic profiling of biological fluids, has been proposed for early and fast diagnosis.
Objectives
To confirm the MIR spectroscopy to discriminate SF samples from patients with septic arthritis from other causes of joint effusion.
Methods
Synovial fluids from 402 patients referred for suspected arthropathies were prospectively collected in six hospitals and stored at °80°C. The infrared absorption spectrum was acquired for each of the frozen samples using a chalcogenide fiber biosensor. The most informative spectral variables were selected and then used to develop an algorithm. Then, the algorithm has been validated on independent synovial fluids collected straight after arthrocentesis from 86 patients.
Results
The calibration (n=402) and validation (n=86) cohorts consists of synovial fluid samples from patients exhibiting various etiologies. These samples (n=488), by using SF bacteriological analysis and culture and 16S PCR analysis were classified as septic arthritis (n=43) or non-septic arthritis (n=443). On the calibration cohort, the performances of the algorithm show a sensitivity of 90%, a specificity of 90%, a NPV of 99% and a PPV of 41%, the area under the ROC curve (AUROC) was 0.95. On the validation cohort, the performances of the algorithm show a sensitivity of 92%, a specificity of 81%, a NPV of 98% and a PPV of 46%, the area under the ROC curve (AUROC) was 0.90.
Conclusions
This study confirms the diagnostic performances of MIR spectroscopy for the discrimination between septic and non-septic synovial fluids. The high negative predictive value and the very short time (within ten minutes) required to obtain the result makes it possible to quickly rule out an infection diagnosis.
