Comparing the use of individual and composite terms to evaluate adverse drug event disproportionality: a focus on glucagon-like peptide-1 receptor agonists and diabetic retinopathy

Evaluation of four machine learning models for signal detection

Background

Logistic regression-based signal detection algorithms have benefits over disproportionality analysis due to their ability to handle potential confounders and masking factors. Feature exploration and developing alternative machine learning algorithms can further strengthen signal detection.

Objectives

Our objective was to compare the signal detection performance of logistic regression, gradient-boosted trees, random forest and support vector machine models utilizing Food and Drug Administration adverse event reporting system data.

Design

Cross-sectional study.

Methods

The quarterly data extract files from 1 October 2017 through 31 December 2020 were downloaded. Due to an imbalanced outcome, two training sets were used: one stratified on the outcome variable and another using Synthetic Minority Oversampling Technique (SMOTE). A crude model and a model with tuned hyperparameters were developed for each algorithm. Model performance was compared against a reference set using accuracy, precision, F1 score, recall, the receiver operating characteristic area under the curve (ROCAUC), and the precision-recall curve area under the curve (PRCAUC).

Results

Models trained on the balanced training set had higher accuracy, F1 score and recall compared to models trained on the SMOTE training set. When using the balanced training set, logistic regression, gradient-boosted trees, random forest and support vector machine models obtained similar performance evaluation metrics. The gradient-boosted trees hyperparameter tuned model had the highest ROCAUC (0.646) and the random forest crude model had the highest PRCAUC (0.839) when using the balanced training set.

Conclusion

All models trained on the balanced training set performed similarly. Logistic regression models had higher accuracy, precision and recall. Logistic regression, random forest and gradient-boosted trees hyperparameter tuned models had a PRCAUC ⩾ 0.8. All models had an ROCAUC ⩾ 0.5. Including both disproportionality analysis results and additional case report information in models resulted in higher performance evaluation metrics than disproportionality analysis alone.

Exendin-4 promotes retinal ganglion cell survival and function by inhibiting calcium channels in experimental diabetes

Progressive damage of retinal ganglion cells (RGCs) is observed in early diabetic retinopathy. Intracellular Ca2+ overload mediated by Ca2+ influx through voltage-gated Ca2+ channels (VGCCs) is involved in neurodegeneration, whereas glucagon-like peptide-1 (GLP-1) provides neuroprotection. However, whether GLP-1 plays a neuroprotective role in diabetic retinas by modulating VGCCs remains unknown. We found that eye drops of exendin-4, a long-acting GLP-1 receptor (GLP-1R) agonist, prevented the increase of L-type Ca2+ current (ILCa) densities of RGCs induced by 4-week hyperglycemia and promoted RGC survival by suppressing L-type VGCC (L-VGCC) activity in streptozotocin-induced diabetic rats. Moreover, exendin-4-induced suppression of ILCa in RGCs may be mediated by a GLP-1R/Gs/cAMP-PKA/ryanodine/Ca2+/calmodulin/calcineurin/PP1 signaling pathway. Furthermore, exendin-4 functionally improved the light-evoked spiking ability of diabetic RGCs. These results suggest that GLP-1R activation enhances cAMP to PP1 signaling and that PP1 inactivates L-VGCCs by dephosphorylating them, thereby reducing Ca2+ influx, which could protect RGCs against excitotoxic Ca2+ overload.

Performance of subgrouped proportional reporting ratios in the US Food and Drug Administration (FDA) adverse event reporting system

BACKGROUND

Many signal detection algorithms give the same weight to information from all products and patients, which may result in signals being masked or false positives being flagged as potential signals. Subgrouped analysis can be used to help correct for this.

RESEARCH DESIGN AND METHODS

The publicly available US Food and Drug Administration Adverse Event Reporting System quarterly data extract files from 1 January 2015 through 30 September 2017 were utilized. A proportional reporting ratio (PRR) analysis subgrouped by either age, sex, ADE report type, seriousness of ADE, or reporter was compared to the crude PRR analysis using sensitivity, specificity, precision, and c-statistic.

RESULTS

Subgrouping by age (n = 78, 34.5% increase), sex (n = 67, 15.5% increase), and reporter (n = 64, 10.3% increase) identified more signals than the crude analysis. Subgrouping by either age or sex increased both the sensitivity and precision. Subgrouping by report type or seriousness resulted in fewer signals (n = 50, -13.8% for both). Subgrouped analyses had higher c-statistic values, with age having the highest (0.468).

CONCLUSIONS

Subgrouping by either age or sex produced more signals with higher sensitivity and precision than the crude PRR analysis. Subgrouping by these variables can unmask potentially important associations.

Relevance of Peptide Homeostasis in Metabolic Retinal Degenerative Disorders: Curative Potential in Genetically Modified Mice

The mammalian retina contains approximately 30 neuropeptides that are synthetized by different neuronal cell populations, glia, and the pigmented epithelium. The presence of these neuropeptides leaves a mark on normal retinal molecular processes and physiology, and they are also crucial in fighting various pathologies (e.g., diabetic retinopathy, ischemia, age-related pathologies, glaucoma) because of their protective abilities. Retinal pathologies of different origin (metabolic, genetic) are extensively investigated by genetically manipulated in vivo mouse models that help us gain a better understanding of the molecular background of these pathomechanisms. These models offer opportunities to manipulate gene expression in different cell types to help reveal their roles in the preservation of retinal health or identify malfunction during diseases. In order to assess the current status of transgenic technologies available, we have conducted a literature survey focused on retinal disorders of metabolic origin, zooming in on the role of retinal neuropeptides in diabetic retinopathy and ischemia. First, we identified those neuropeptides that are most relevant to retinal pathologies in humans and the two clinically most relevant models, mice and rats. Then we continued our analysis with metabolic disorders, examining neuropeptide-related pathways leading to systemic or cellular damage and rescue. Last but not least, we reviewed the available literature on genetically modified mouse strains to understand how the manipulation of a single element of any given pathway (e.g., signal molecules, receptors, intracellular signaling pathways) could lead either to the worsening of disease conditions or, more frequently, to substantial improvements in retinal health. Most attention was given to studies which reported successful intervention against specific disorders. For these experiments, a detailed evaluation will be given and the possible role of converging intracellular pathways will be discussed. Using these converging intracellular pathways, curative effects of peptides could potentially be utilized in fighting metabolic retinal disorders.

The association between GLP-1 receptor agonist and diabetic ketoacidosis in the FDA adverse event reporting system

BACKGROUND AND AIMS

In 2019, the Medicines and Healthcare products Regulatory Agency (MHRA) of the United Kingdom (UK) and food and drug administration (FDA) of the United States of America (US) suggested that the relationship between glucagon-like peptide-1 receptor agonists (GLP-1RA) and diabetic ketoacidosis (DKA) deserved attention. This study is aiming to assess the association between GLP-1RA and DKA/ketosis in the FDA Adverse Event Reporting System (FAERS) database.

METHODS AND RESULTS

Using FAERS database, we firstly extract the number of DKA reports from the first quarter (Q1) of 2004 to the fourth quarter (Q4) of 2019 and calculate proportional reporting ratios (PRRs). We then mined each FAERS file from 2004 Q1 to 2020 Q4 and obtained detailed information on DKA reports. From the first quarter (Q1) of 2004 to the fourth quarter (Q4) of 2019, there are 1382 DKA cases (1491 ketosis cases) associated with GLP-1RA in the FAERS database. There was a slight disproportionate reporting of DKA associated with overall GLP-1RA (PRR 1.49, 95%CI 1.24-1.79, p < 0.001) after excluding the impact of SGLT2i, T1D and insulin. Any disproportionality disappeared after selecting the GLP-1RA combined with insulin for comparison.

CONCLUSIONS

When GLP-1RA not combined with insulin, the disproportionality of DKA reports associated with GLP-1RA was observed. Our analysis mined the FAERS database to provide evidence and highlight the potential association between DKA adverse events and GLP-1RA therapy that clinicians tend to overlook.

Semaglutide and Diabetic Retinopathy Risk in Patients with Type 2 Diabetes Mellitus: A Meta-Analysis of Randomized Controlled Trials

BACKGROUND

Semaglutide is a recently approved glucagon-like peptide-1 receptor agonist used to treat patients with type 2 diabetes mellitus (T2DM). The SUSTAIN 6 trial found a significantly higher rate of retinopathy complications in the semaglutide-treated group compared with the placebo group.

OBJECTIVE

This study aimed to evaluate the association between semaglutide and the risk of retinopathy in patients with T2DM.

METHODS

Electronic databases were systematically searched up to April 2021 to identify randomized controlled trials that reported diabetic retinopathy (DR) events in semaglutide-treated and control groups. A meta-analysis was conducted using Review Manager 5.4 software to calculate the risk ratio (RR) and 95% confidence intervals (CIs).

RESULTS

A total of 23 randomized trials involving 22,096 patients with T2DM were included. There were 730 incident DR cases-463 in the semaglutide group and 267 in the control group. Overall, semaglutide was not associated with increased DR risk compared with controls when all trials were combined (RR 1.14, 95% CI 0.98-1.33). Subgroup analysis showed that semaglutide was associated with an increased risk of DR compared with placebo (RR 1.24, 95% CI 1.03-1.50). Moreover, patient age ≥ 60 years and diabetes duration ≥ 10 years were also factors for increased risk of DR when using semaglutide (RR 1.27, 95% CI 1.02-1.59; RR 1.28, 95% CI 1.04-1.58, respectively).

CONCLUSIONS

Semaglutide was not associated with an increased risk of DR; however, caution regarding DR risk is needed for older patients or those with long diabetes duration when taking semaglutide.