Biguanides

The activity of two drugs currently used against Acanthamoeba was tested on Acanthamoeba griffini. Chlorhexidine (Chlorhexidine digluconate, Alfa Aesar, Germany) is a standard antiseptic belonging to the biguanides family which are commonly used in contact lens maintenance solutions; and voriconazole (Sigma, Madrid, Spain) an inhibitor of ergosterol synthesis that has been proven previously to be highly effective against clinical strains of Acanthamoeba [22 (link), 23 ]. For the sensitivity and activity assays, a type strain from the American Type Culture Collection (ATCC), Acanthamoeba castellanii Neff ATCC 30010, genotype T4 was used as a control.
For the activity assays a previously developed colorimetric 96-well microtiter plate assay, based on the oxide-reduction of Alamar Blue assay [24 (link)], was used for the determination of drug efficacy against the trophozoites of the selected Acanthamoeba strains. Subsequently the plates were analyzed, over a period from 72 to 120 h, on a Microplate Reader Model 680 (Biorad, Hercules, CA) using a test wavelength of 570 nm and a reference wavelength of 630 nm. For those strains that were sensitive to the assayed drugs, the percentage of inhibition and 50% inhibitory concentrations (IC₅₀) were calculated by linear regression analysis using a 95% confidence limit. All experiments were performed three times each in duplicate and the mean values were also calculated. A paired two-tailed t-test was used for analysis of the data. Values of P < 0.05 were considered significant. The inhibition curves of the statistical analysis were developed using the Sigma Plot 12.0 software programme (Systat Software Inc.).

We searched the database of patients with type 2 diabetes who were referred for ECG-gated coronary computed tomography angiography (CCTA) examinations (Toshiba Aquilion CT scanner, Toshiba Medical, Tochigi, Japan; SOMATOM Definition or Force, Siemens Healthineers, Forchheim, Germany) for the first time and underwent abdominal CT scans (Toshiba Aquilion CT scanner, Toshiba Medical, Tochigi, Japan; Discovery CT750 HD, General Electric Healthcare, Chicago, IL, USA; SOMATOM Definition, Siemens Healthineers, Forchheim, Germany; GE Optima CT660, General Electric Healthcare, Chicago, IL, USA) within 1 year of CCTA at Osaka University Hospital or Sumitomo Hospital between January 2000 and March 2021. A total of 411 patients met these criteria. Among these patients, we excluded patients to avoid the influence of cardiac function or the myocardial CT value of pathological conditions other than myocardial fat accumulation. The excluded patients included those with heart failure with reduced ejection fraction (≤ 40%), those with valvular heart disease and those who had received past percutaneous coronary intervention (PCI) for coronary artery disease. Moreover, we excluded patients who had liver cirrhosis, renal failure (estimated glomerular filtration rate of < 30 mL/min/1.73 m²), malignant diseases and diseases requiring glucocorticoids for the treatment of other diseases. Furthermore, it is known that changes in X-ray tube voltage affect CT attenuation values [20 (link)]. Therefore, patients who underwent CCTA or abdominal CT examinations that were not performed at 120 kV were excluded. Using these criteria, 124 patients were finally included in our analyses. The flowchart for the recruitment of the patients is shown in Additional file 1: Fig S1. Among these 124 patients, the medications for diabetes at the time of CCTA were as follows: insulin for 42 patients, glucagon-like peptide-1 (GLP-1) receptor agonists for 9 patients, sulfonylureas for 31 patients, biguanides for 43 patients, dipeptidyl peptidase-4 inhibitors for 40 patients, α-glucosidase inhibitors for 26 patients, thiazolidinediones for 13 patients, glinides for 12 patients and sodium–glucose cotransporter 2 (SGLT2) inhibitors for 3 patients. The medications for dyslipidemia at the time of CCTA were as follows: statins for 69 patients, fibrates for 9 patients, ezetimibe for 5 patients, omega-3 fatty acids for 8 patients and probucol for 1 patient.
This study was approved by the Institutional Ethics Review Boards of Osaka University Hospital and Sumitomo Hospital and was carried out in accordance with the principles of the Declaration of Helsinki. The study was announced to the public on the websites of our department at Osaka University Hospital and Sumitomo Hospital, and all patients were allowed to participate or refuse to participate in the study.

By using the Oracle Empirica Signal software (Oracle Health Sciences, Austin, TX), we calculated disproportionality statistics produced by four signal detection methodologies, to assess the occurrence of therapy failure (cases) in depressed patients, in association with the exposure to at least one antidiabetic drug, defined as the following ATC Level 4 codes: A10BA Biguanides; A10BB Sulfonylureas; A10BG Thiazolidinediones; A10BH DPP4-inhibitors; A10BJ GLP-1 analogues; A10BK SGLT2 inhibitors (i.e., those agents for which preliminary evidence from literature supports our pharmacological hypothesis).
Three of these disproportionality scores, based on 2 × 2 disproportionality analysis, are well-established and currently used worldwide by several organisations for routine safety surveillance, i.e:

i) The Reporting Odds Ratio (ROR), defined as the ratio of the odds of the occurrence of therapy failure with antidiabetic drugs versus the occurrence of therapy failure without antidiabetic agents (van Manen et al., 2007 (link));

ii) The Proportional Reporting Ratio (PRR), comparing the frequency of occurrence of therapy failure in reports referring to antidiabetic agents with the frequency of occurrence of reports of therapy failure in reports that do not mention antidiabetic agents. (van Manen et al., 2007 (link)).

iii) The Empirical Bayesian Geometric Mean (EBGM) calculated using the Multi-item Gamma Poisson Shrinker (MGPS) Algorithm, using Bayesian shrinkage to improve the reliability of the disproportionality score (DuMouchel, 1999 (link)). We generated both the point estimates (EBGM) and their associated 90% confidence intervals labelled EB05–EB95.

Moreover, we used a more advanced regression-based methodology designed to produce disproportionality statistics with adjusted background rates; it can control masking and more extensive confounding effects by fitting separate Bayesian logistic regression models to each target AE and by automatically selecting predictors to be included in each regression model:

iv) The Regression-enhanced Empirical Bayesian Geometric Mean (ERAM) calculated using the Regression-Adjusted Gamma Poisson Shrinker (RGPS) Algorithm (DuMouchel and Harpaz, 2012 ). We generated the point estimates (ERAM) and their associated 90% confidentiality intervals labelled ER05–ER95.

With the aim to investigate the antidepressant effects of antidiabetic drugs, disproportionality signals were considered clinically meaningful if.

i) The upper limit of the 90% confidence interval (CI) of the ROR for cases (ROR95) is less than one;

ii) The PRR score is less than one and the corresponding p-value is less than 0.05;

iii) The upper limit of the 90% confidence interval of the EBGM for cases (EB95) is less than one;

iv) The upper limit of the 90% confidence interval of the ERAM for cases (ER95) is less than one.

Publicly available FAERS data from 1 January 2004 to 30 September 2021 were downloaded from the FDA website as raw data. Hypoglycemic medications were drugs mapped to the Anatomic Therapeutic Chemical Classification (ATC) as A10 class (antidiabetic drugs, ATCA10), including insulin (including insulin and its analogs discussed in this paper) and SGLT2is, as well as biguanides, dipeptidyl peptidase-4 (DPP4), glinides (GLN), glucagon-like peptide 1 (GLP-1), sulfonylureas (SUs), and thiazolidinediones (TZDs). Osteomyelitis was defined as all of the AEs containing the keyword “osteomyelitis,” which were determined by using the Standardized MedDRA Query (SMQ, version 23.0) terms (Katsuhara and Ikeda, 2021 (link)), within “Osteonecrosis (SMQ).” The dataset of “Diabetes” was composed of all reports in FAERS containing the keyword “diabetes.” The following criteria of exclusion (Figure 1) were applied: each potential case was subjected to a data-cleaning procedure to remove reports that were officially deleted by FDA authority, that were duplicated, with missing case ID and date, or with inaccurate data for gender and age. The obtained reports were then filtered with the targeted drug as the primary suspected (PS) drug. All the reports containing hypoglycemic medications other than the targeted medication were removed, to minimize the possibility of interfering effects.