Experience with the Galactomannan Assay at a Cancer Center

YONHEE CHA, MD, SEJEAN SOHN, MPH, GIANNA ZUCCOTTI, MD, MPH, JEFFREY STILES, BS, TIMOTHY KIEHN, PhD., KENT SEPKOWITZ,

Author address: 

MD; Memorial Sloan-Kettering Cancer Center, New York, NY.

Abstract: 

Background: Diagnosis of invasive aspergillosis (IA) remains difficult to establish. In 2003, the FDA approved the galactomannan antigen (GM) test for serologic detection. To examine its utility, we reviewed our experience with the test in 100 consecutive patients. Methods: The GM test (Platelia Aspergillus; Bio-Rad Laboratories) became available at MSKCC 108722;03. GM was ordered at the discretion of the primary MD. Charts were abstracted for pertinent clinical information and patients assigned to groups defined by 2002 Invasive Fungal Infections Cooperative Group (IFICG) consensus classification for IA. GM result was not used to characterize patients. Results: 100 patients (35 leukemia/lymphoma, 53 transplant, 12 other) had 333 GM tests from 108722;1- 03 to 38722;1-04. By IFICG criteria, 7 had proven; 4 had probable; 36 had possible, and 53 had no evidence of IA. 14 had GM+, including 2/7 proven and 2/4 probable, 9 possible and 1 without evidence of IA. Most GM+ cases had received antimold therapy pre-test (range 0 to >45 days, median 9 days). 6 of the 7 GM8722; proven/probable cases were on antimold therapy a median 21 days (range 2 to >45 days) before GM was sent; 1 GM8722; probable case was not on antimold therapy. Proven-probable GM+ cases had significantly higher median GM index than possible/no evidence cases (1.238 vs 0.752, p=0.03). Overall, 29 patients had or were receiving piperacillin-tazobactam (P-T) when tested for GM including 13 of 14 GM+ patients. GM for proven/probable cases was 36% sensitive. Examining 7 proven and 53 no-evidence groups only, the positive predictive value (PPV) was 67% and negative predictive value (NPV) was 91%. Conclusion: The GM test was insensitive for detecting proven/probable IA but had adequate PPV and NPV when examining "œbest" cases. The confounding effects of pre-test antimold therapy and P-T appear substantial and require further study. Given these results, we are reluctant to adopt routine GM testing of at-risk patients at our hospital.
2004

abstract No: 

NULL

Full conference title: 

42nd Annual Meeting Infectious Diseases Society of America
    • Infectious Diseases Society of America 42nd