Assessment of clinical performance without adequate analytical validation: A prescription for confusion.

OBJECTIVES: Lp-PLA2 is a biomarker with promise for predicting cardiac risk. The lack of reproducible results has limited its use. In evaluating a new reagent kit, we investigated conditions for optimal reproducibility. METHODS: The Auto-PLAC reagents were evaluated on the Cobas instrument. Performance characteristics, stability, and population ranges were determined. RESULTS: Analytical performance characteristics replicated manufacturer's claims. The stability profile of the analyte was unusual, with increasing results observed with storage at 4°C or -20°C. Only storage at -70°C gave acceptable stability. Population median values with properly preserved samples were much lower than the cut off previously validated for increased risk. CONCLUSIONS: It is postulated that variability in specimen handling was a major contributor to the lack of traceability of the current reagents to the earlier clinical studies demonstrating its utility. We are now unsure how to identify reliable criteria for result interpretation.

Insulin autoimmunity and hypoglycemia in seven white patients.

OBJECTIVE: To report the clinical, biochemical, and immunologic characteristics of 7 white patients with the rare disorder of hyperinsulinemic hypoglycemia in association with spontaneously generated high titers of antibodies to human insulin. METHODS: We reviewed the clinical data, history, and symptoms of the 7 study patients and summarized the biochemical findings during a spontaneous episode of hypoglycemia. Insulin antibody binding was measured in all patients, and antibody affinity, capacity, and clonality were analyzed in 4. A mixed meal study was conducted in 2 patients. A potential mechanism for postprandial hypoglycemia is presented. RESULTS: In all 7 patients (6 women and 1 man), symptoms were neuroglycopenic, occurring primarily postprandially but during fasting in some patients. During hypoglycemia, concentrations of insulin, proinsulin, and, in most patients, C peptide considerably exceeded those observed in patients with insulinoma. These concentrations were spuriously elevated as a result of interference by the autoantibodies in the immunoassays. No patient had evidence of an insulinoma on various radiologic localization procedures directed at the pancreas. Insulin antibodies showed a high percentage of binding to human insulin--50 to 90%. Heterogeneity of antibodies regarding clonality and antibody binding sites was noted; some patients had polyclonal and some had monoclonal IgG class antibodies. Most patients had two categories of binding sites: high affinity/low capacity and low capacity/high affinity. Although the mechanism for postprandial hypoglycemia remains conjectural, prolonged elevations of postprandial concentrations of total and free insulin are consistent with the putative mechanism of a buffering effect of insulin antibodies. CONCLUSION: Insulin autoimmune hypoglycemia, although rare in any racial group and especially in white subjects, can be readily detected by high titers of insulin antibodies. Such a determination should be done in all patients undergoing evaluation for hypoglycemia.