Reliability of cold agglutinin test (CAT) for the detection of patients with Mycoplasma pneumoniae pneumonia

Introduction:
M. pneumoniae is one of the causative agents of primary atypical pneumonia.  This infection causes 20-40% of community acquired pneumonia and is associated with an array of extra-pulmonary manifestations. There is a need for a rapid diagnostic test in order to prescribe prompt and appropriate antibiotic therapy. Even though isotype specific antibody testing provides definitive diagnosis, paired sera testing does not help in real time diagnosis. Cold agglutinins detectable by the Cold Agglutination Test (CAT) appear and disappear early in infection compared to long lasting specific antibodies that are detectable by specific immunoassays. Although there are some reports suggesting CAT is unreliable, it is being often used to diagnose M. pneumoniae pneumonia in Sri Lankan clinical settings. The aim of the current study was to  evaluate the use of CAT as a bed-side screening test for early diagnosis of M. pneumoniae pneumonia compared to ELISA for detection of specific antibodies in the Sri Lankan context.

Methods:
Ninety seven clinically and radiologically confirmed patients with pneumonia were enrolled in the study. CAT was performed on acute stage sera. A CAT titer ≥1/32 was considered as positive. Isotype specific M. pneumoniae ELISA with paired sera was compared with CAT results.

Results:
Mycoplasma pneumonia was confirmed in 15 of the 97 patients in the study using Mycoplasma specific IgM and 4 fold rise in titre. Of these, 3 were positive by the CAT. The sensitivity and specificity of the CAT compared to IgM/4fold rise in IgG detection were 20% (3/15) and 81.7% (67/82) respectively.  Negative and positive predictive values of the CAT compared to ELISA were 84.8% (67/79) and 16.7% (3/18) respectively.

Conclusion:
CAT is not a reliable screening test compared to specific antibody detection by isotype ELISA for the detection of M. pneumoniae pneumonia due to its low sensitivity and positive predictive values.