Adarsh Varma1, Sheri Trudeau2, Yueren Zhou2, Syed-Mohammed Jafri1, Richard Krajenta2, Lois Lamerato2, Kimberly Brown1, Veronica Luzzi3, Mei Lu2, Stuart C Gordon4. 1. Department of Gastroenterology and Hepatology, Henry Ford Health System, 3E One Ford Place, Detroit, MI, 48202, USA. 2. Department of Public Health Sciences, Henry Ford Health System, Detroit, MI, USA. 3. Providence Health and Services Regional Laboratory, Portland, OR, USA. 4. Department of Gastroenterology and Hepatology, Henry Ford Health System, 3E One Ford Place, Detroit, MI, 48202, USA. sgordon3@hfhs.org.
Abstract
BACKGROUND AND AIMS: Normal ranges of serum alanine aminotransferase (ALT) may vary by race. However, results from research studies are contradictory, and many of these studies have included only small numbers of African Americans. We investigated ALT values in patients without evidence of liver disease to determine whether normal ranges differ across race groups. We also evaluated whether a race- and sex-dependent upper limit of normal (ULN) would improve the ability of ALT to predict liver disease compared to the sex-dependent ULN currently in use. METHODS: We identified ICD9 codes for liver conditions and diabetes in medical records from a sample of 6719 patients. Analysis of variance (ANOVA) was used to assess differences in ALT log-transformed distributions by race. Logistic regression was used to evaluate whether the addition of race to the current sex-dependent ULN improves the ability of ALT to predict liver disease (assessed by area under the receiver operating characteristic curve (AUROC)). RESULTS: Among 1200 patients with BMI 18.5 < 25 and no evidence of liver disease or type 2 diabetes in their medical record, African Americans demonstrated significantly lower ALT (23.47 IU/L; 95% CL 22.87-24.10) than a combined group of Asian American/White/Other patients (25.71 IU/L; 95% CL 24.69-26.77). This difference remained across BMI categories. The race- and sex-dependent model demonstrated significantly better predictive ability than the sex-dependent model (AUROC = 66.6% versus 59.6%, respectively; p < 0.0001). CONCLUSIONS: In a large, racially diverse sample, African Americans demonstrated significantly lower ALT compared to non-African Americans; this difference remained as BMI increased. The establishment of race-specific normal ranges for ALT could contribute to better screening and care for African American patients.
BACKGROUND AND AIMS: Normal ranges of serum alanine aminotransferase (ALT) may vary by race. However, results from research studies are contradictory, and many of these studies have included only small numbers of African Americans. We investigated ALT values in patients without evidence of liver disease to determine whether normal ranges differ across race groups. We also evaluated whether a race- and sex-dependent upper limit of normal (ULN) would improve the ability of ALT to predict liver disease compared to the sex-dependent ULN currently in use. METHODS: We identified ICD9 codes for liver conditions and diabetes in medical records from a sample of 6719 patients. Analysis of variance (ANOVA) was used to assess differences in ALT log-transformed distributions by race. Logistic regression was used to evaluate whether the addition of race to the current sex-dependent ULN improves the ability of ALT to predict liver disease (assessed by area under the receiver operating characteristic curve (AUROC)). RESULTS: Among 1200 patients with BMI 18.5 < 25 and no evidence of liver disease or type 2 diabetes in their medical record, African Americans demonstrated significantly lower ALT (23.47 IU/L; 95% CL 22.87-24.10) than a combined group of Asian American/White/Other patients (25.71 IU/L; 95% CL 24.69-26.77). This difference remained across BMI categories. The race- and sex-dependent model demonstrated significantly better predictive ability than the sex-dependent model (AUROC = 66.6% versus 59.6%, respectively; p < 0.0001). CONCLUSIONS: In a large, racially diverse sample, African Americans demonstrated significantly lower ALT compared to non-African Americans; this difference remained as BMI increased. The establishment of race-specific normal ranges for ALT could contribute to better screening and care for African American patients.