We would like to thank Paul etal. for their interest in our article.1 Here, they report on a predicted positive relationship between body mass index (BMI), serum ferritin (SF), and transient elastography score (TES) in chronic hepatitis B patients using regression modeling. Furthermore, using interaction regression models, they found that, among subjects with BMI >28 kg/m2 increased SF was associated with increased TES, including a higher probability of TES above the cirrhosis cutpoint of >13 when SF levels were >650 µg/L.
In our study, we found that both BMI and SF >1.5 times the upper limit of normal (ULN) were independent predictors of advanced fibrosis using multivariate stepwise logistic regression analysis, including the following variables, selected a priori, known to be associated with severity of nonalcoholic fatty liver disease: age at biopsy; sex; BMI; presence of diabetes; and alanine aminotransferase. Despite finding that both BMI and SF >1.5 times the ULN independently predicted advanced fibrosis, when we conducted additional interaction regression analyses with regard to BMI, SF, and advanced fibrosis to see whether similar relationships were also evident in our cohort, we did not find a significant interaction of BMI and SF on advanced fibrosis. In fact, we found a significantly lower BMI in subjects above SF >1.5 versus SF ≤1.5 times the ULN (32.6 ± 5.5 versus 34.4 ± 6.5; P = 0.004). The reason for this lies within our study cohort; our cohort consisted of 141 premenopausal women who had a significantly higher mean BMI (36.0 kg/m2) than the rest of the cohort, but a lower SF level (139 ng/mL) as a result of ongoing menses (Table 1). Furthermore, this group accounted for only 12% of subjects with advanced (stage 3 and 4) fibrosis. Thus, in analyzing our total cohort for a statistical interaction of BMI/SF on fibrosis, this subgroup of premenopausal women, with a higher mean BMI, likely biased any potential positive effects of BMI and SF upon fibrosis. To investigate this further, when we repeated the interactive regression models individually in each group shown in Table 1 we did find a small, but significant, interaction of SF and BMI in male subjects alone (odds ratio: 1.0003; P = 0.028). We would like to also note that there are quite a few differences between our cohort and that of Paul etal.; in addition to being different diseases, our cohort had fewer males (37% versus 54%), higher BMI (34.6 versus 25.4), and higher SF (males, 263 versus 205 ng/mL; females, 110 versus 88 ng/mL).
|No. (%)||237 (37.7)||250 (39.8)||141 (22.4)||628 (100)|
|BMI, kg/m2||33.1 ± 5.4||33.8 ± 6.4||36.0 ± 7.3||34.0 ± 6.3|
|SF, ng/mL||356 ± 337||222 ± 275||139 ± 156||254 ± 293|
|Presence of advanced fibrosis, no. (%)||47 (29)||94 (58)||20 (12)||161 (26)|