University of Michigan
Department of Pediatrics
Evidence-Based Pediatrics Web Site

Question

• How good is acanthosis nigricans at identifying children at risk for hyperinsulinemia?

Clinical Bottom Lines

1. Acanthosis nigricans (AN) is strongly associated with hyperinsulinemia, but it loses added diagnostic value when a BMI greater than 3 SD is considered first. The +LR for acanthosis in the general population is 125, but in the setting of obesity its +LR is only 8.
   
2. The value of (AN) as a marker for hyperinsulinemia changes depending on the population studied. Specifically, it is important to know that the sensitivity and specificity of tests (not just prevalence) changes with a change in clinical setting. In this case, specificity slightly decreases because of an increase in false positives in a referral setting, but the +LR decreases more markedly.

Summary of Key Evidence

1. In 139 obese children age 6-10, 35 were identified as having insulin resistance, defined as having in insulin level greater than 15 microunits/mL.¹
2. Only 17 (48%) of those with insulin resistance had AN. 6 children had AN, but normal fasting insulin levels. In this population, the +LR for AN was 8 (only moderately helpful, but of no added value when obesity considered first).

Additional Comments

• This study looked only at obese children. In the general (your clinic) population, if you assume that it is rare for non-obese children age 6-10 to have acanthosis nigricans, then the specificity approaches 100% (99.6), with a positive likelihood ratio of 125 (very helpful). But you would still miss half of the insulin resistant kids if you ONLY used AN to try to identify them. (sensitivity 50%)
  
• The above statement assumes an obesity prevalence of 10% and insulin resistance of 25% within that population. The proportion of insulin resistant children may be higher in extremely obese children.²
  
• If you use obesity as the marker for insulin resistance and make some similar assumptions (the above assumptions plus an assumption that non-obese insulin-resistant children are rare (less than 3/1000), then obesity is 92.5% sensitive and 92% specific for insulin resistance. This is much better for identifying ALL the children you may wish to screen for insulin resistance and, presumably, the prediabetic state.
  

Citation

1. Nguyen T, et al. Relation of acathosis nigricans to hyperinsulinemia and insulin sensitivity in overweight African American and white children. J Pediatrics 2001; 139:474-80.
   
2. Sinha, et al. Prevalence of impaired glucose tolerance among children and adolescents with marked obesity. N Engl J Med 2002; 346(11):802-10.
   
3. Sackett DL. The architecture of diagnostic reasearch. BMJ 2002; 324:539-541.
   

CAT Author: Kimberly J. Stewart, MD

CAT Appraisers: Robert Schumacher, MD

Date appraised: April 1, 2002