Acanthosis Nigricans and Nephropathy in Diabetes Mellitus: Is the Association Causal?

Dear Editor,

I read with interest the study by Dutta et al. exploring the relationship between acanthosis nigricans (AN) and type 2 diabetes mellitus. Though the study is interesting, I would like to highlight certain issues regarding the methodology that have a bearing on the conclusions [Supplementary Appendix].¹ Although the study design is labeled “prospective,” it is in fact cross-sectional because the “exposure” (AN) and “outcome” (nephropathy) were measured at the same time. This distinction is important because cross-sectional studies cannot establish a causal relationship.² The use of terms like “predicts” is not appropriate for the results of a cross-sectional study. There is a discrepancy in the calculation using a 2*2 contingency table. The unadjusted odds ratio (OR) should be the same when exposure and outcome are reversed.³ But the authors report ORs of 4.77 at one place and 4.93 at another. The authors have not accounted for a few obvious confounders. For example, semaglutide was used for 22.7% of subjects in the AN group and only 6.7% in the NAN group. This was not adjusted in the multivariate model. Next, KDIGO defines CKD based on the albumin creatinine ratio, which is usually in mg/g units. But the authors have used only albuminuria in mg/L. Moreover, albuminuria is expressed as mean and SD. But SD is more than three times the mean. This clearly shows that the data is skewed. For skewed data, using parametric tests is inappropriate. Nonparametric tests such as the Mann-Whitney test are recommended for skewed data. eGFR is not statistically different between the two groups. The observed difference in CKD prevalence between the groups is entirely driven by albuminuria. So, if albuminuria is not measured, the conclusions become unreliable. The authors claim that age-matching was performed. But the methodology is unclear about the matching method. Also, some exclusion criteria are unclear. The authors state they excluded congestive heart failure, yet heart failure is included as a macrovascular outcome. This seriously affects the internal validity.

Given all the above concerns, the results of this cross-sectional study are hypothesis-generating. A longitudinal cohort design to identify incident CKD, with adjustment for important confounders, will be necessary to establish a strong association and causality.^4,5