We thank Drs. Cao and Li for their comment (1) on our article (2). We also think that the tools available to check the risk of any rare adverse event, such as worsening type 2 diabetes or new-onset diabetes, lack precision. In our case, a large set of evidence and the potential clinical utility of therapy with proprotein convertase subtilisin/kexin type 9 inhibitors (PCSK9i) provide plausibility for the investigation and relevance for the report.
A major limitation that we had to deal with in our meta-analysis is the fact that PCSK9i studies were not designed for testing the hypothesis and hence no information was available on the concomitant use and titration of antidiabetes medicines up or down. This drawback may certainly overshadow the sizing of the effect as well as its statistical significance (2). To partially overcome this issue, we separated individuals with and without diabetes at study entry and found very similar results in both groups, reducing our concern about this potential bias.
As observed by Cao and Li (1), we selected all the available clinical trials on PCSK9i, irrespective of market withdrawal. The reason for this choice was our expectation of a small size effect based on genetic studies and the short duration of clinical trials included in the meta-analysis. Therefore, to avoid restricting statistical accuracy, we included all data available in studies using PCSK9i. Indeed, the probability of type II error increased 48 times (from 0.7% to 33.8%) when SPIRE trials were removed. From the mechanistic viewpoint, as any possible unbalance in glucose homeostasis would not result from the antibody per se but from the reduction of plasma PCSK9, we did not consider it pertinent to exclude or discriminate the effects by the degree of homology to the human antibody.
We based our hypothesis on the mechanistic assumption that, by improving LDL receptor (LDLR) turnover, PCSK9i would potentially decrease pancreatic β-cell function and insulin sensitivity (3). This concept is based on a set of evidence: 1) the inverse association between the severity of LDLR dysfunction in patients with familial hypercholesterolemia and the propensity to develop type 2 diabetes (4); 2) LDLR role of regulating pancreatic β-cell homeostasis in in vitro and in vivo models (3,4); 3) isolated rat islet β-cell death reduction by LDL coincubation, an LDLR-dependent effect (4); 4) glucose-stimulated insulin secretion decreases by LDL in rodent pancreatic islets, a mechanism abolished in LDLR knockout animals (4); and 5) gene-mediated changes in plasma cholesterol influenced by PCSK9, LDLR, HMG-CoA reductase, Niemann-Pick-C1-Like-1, and ABCG5/G8 pathways are equally related to the incidence of type 2 diabetes (5).
Our results on fasting blood glucose and HbA1c change after PCSK9i point toward the same direction of in vivo, in vitro, and Mendelian randomization studies. Altogether, this evidence supports the concept of an inverse and dose-dependent effect of plasma LDL on glucose homeostasis. As with statins, treatment with PCSK9i had a significant but minimal effect on glycemic levels. Many questions remain unanswered, such as the persistence of this effect in the long term or the predisposing factors for the effect to occur. Nevertheless, this mild effect has no power to eclipse the proven clinical benefit observed in randomized clinical trials of both statins and PCSK9i therapies.
Article Information
Funding. A.C.S. was supported by a fellowship grant of research productivity from Brazilian National Research Council (CNPq).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
