I read with interest the article by Bambha et al.1 examining the role of ethnicity in nonalcoholic fatty liver disease (NAFLD). Using the Nonalcoholic Steatohepatitis (NASH) Clinical Research Network database, the authors examined associations between ethnicity and NAFLD with greater scrutiny than has previously been published. There are challenges, however, in separating the role of ethnicity from other risk factors for NAFLD. Two potential confounding variables were underemphasized by the authors.
Age has been confirmed as a risk factor for fibrosis in NASH time and again.2 The age difference between Latinos and non-Latino whites in this study (44.2 versus 50.9, P < 0.0001) may have resulted in the assessment of these patients at different stages of disease progression. This is highlighted by the finding that age was independently associated with advanced fibrosis and ethnicity was not. The authors acknowledged this limitation and accounted for age in the logistic regression analysis. However, this difference makes the groups inhomogeneous and may influence other comparisons between them. As reported, Latinos had more grade ≥2 lobular inflammation (61% versus 48%, P = 0.008) and less stage >2 fibrosis (23% versus 30%, P = 0.004) than non-Latino whites. How many of the Latino patients with inflammation would have developed fibrosis if they had been studied, on average, 6.7 years later?
The authors demonstrated that homeostasis model assessment-insulin resistance (HOMA-IR) was independently associated with NASH in non-Latino whites but not in Latinos, despite the fact that HOMA-IR scores for these groups did not differ (5.1 versus 4.6, P = 0.55). Because the frequency of biopsy-proven NASH in Latinos and non-Latino whites was the same (63% versus 62%, no P value), there must be additional risk factors that are more prevalent in Latinos. The authors acknowledged that the PNPLA3 rs738409 polymorphism—more prevalent in Latinos than non-Latino whites3—may be such a risk factor but underemphasize its potential importance. Speliotes et al. recently showed that the rs738409 G allele showed an independent association with histologic measures of NAFLD but not with metabolic traits.4 Thus PNPLA3 polymorphism status may act as a confounder for potential associations between ethnicity and NAFLD, and must be accounted for in future analyses.
With this meticulous analysis, Bambha et al. set a precedent for future studies looking at the role of ethnicity in the progression of NAFLD. But toward the goal of understanding the complex relationships among the many risk factors for this disease, there is still much work to do.