Deep Learning Models Compared to Experimental Variability for the Prediction of CYP3A4 Time-Dependent Inhibition

Most drugs are mainly metabolized by cytochrome P450 (CYP450), which can lead to drug-drug interactions (DDI). Specifically, time-dependent inhibition (TDI) of CYP3A4 isoenzyme has been associated with clinically relevant DDI. To overcome potential DDI issues, high-throughput in vitro assays were established to assess the TDI of CYP3A4 during the discovery and lead optimization phases. However, in silico machine learning models would enable an earlier and larger-scale assessment of TDI potential liabilities. For CYP inhibition, most modeling efforts have focused on highly imbalanced and small data sets. Moreover, assay variability is rarely considered, which is key to understand the model's quality and suitability for decision-making. In this work, machine learning models were built for the prediction of TDI of CYP3A4, evaluated prospectively, and compared to the variability of the experimental assay. Different modeling strategies were investigated to assess their influence on the model's performance. Through multitask learning, additional data sets were leveraged for model building, coming from public databases, in-house CYP-related assays, or other pharmaceutical companies (federated learning). Apart from the numerical prediction of inactivation rates of CYP3A4 TDI, three-class predictions were carried out, giving a negative (inactivation rate kobs < 0.01 min-1), weak positive (0.01 ≤ kobs ≤ 0.025 min-1), or positive (kobs > 0.025 min-1) output. The final multitask graph neural network model achieved misclassification rates of 8 and 7% for positive and negative TDI, respectively. Importantly, the presented deep learning-based predictions had a similar precision to the reproducibility of in vitro experiments and thus offered great opportunities for drug design, early derisk of DDI potential, and selection of experiments. To facilitate CYP inhibition modeling efforts in the public domain, the developed model was used to annotate ∼16 000 publicly available structures, and a surrogate data set is shared as Supporting Information.

Adapting Deep Learning QSPR Models to Specific Drug Discovery Projects

Medicinal chemistry and drug design efforts can be assisted by machine learning (ML) models that relate the molecular structure to compound properties. Such quantitative structure-property relationship models are generally trained on large data sets that include diverse chemical series (global models). In the pharmaceutical industry, these ML global models are available across discovery projects as an "out-of-the-box" solution to assist in drug design, synthesis prioritization, and experiment selection. However, drug discovery projects typically focus on confined parts of the chemical space (e.g., chemical series), where global models might not be applicable. Local ML models are sometimes generated to focus on specific projects or series. Herein, ML-based global models, local models, and hybrid global-local strategies were benchmarked. Analyses were done for more than 300 drug discovery projects at Novartis and ten absorption, distribution, metabolism, and excretion (ADME) assays. In this work, hybrid global-local strategies based on transfer learning approaches were proposed to leverage both historical ADME data (global) and project-specific data (local) to adapt model predictions. Fine-tuning a pretrained global ML model (used for weights' initialization, WI) was the top-performing method. Average improvements of mean absolute errors across all assays were 16% and 27% compared with global and local models, respectively. Interestingly, when the effect of training set size was analyzed, WI fine-tuning was found to be successful even in low-data scenarios (e.g., ∼10 molecules per project). Taken together, this work highlights the potential of domain adaptation in the field of molecular property predictions to refine existing pretrained models on a new compound data distribution.

Flying high?-Exploring the effect of red bull on blood pressure after microsurgical breast reconstruction in a randomized controlled trial

BACKGROUND

A sufficiently high blood pressure (BP) is essential for flap perfusion after microsurgical breast reconstruction. However, postoperative hypotension is common after these procedures. Perioperative volume overload may increase flap-related complications, and postoperative vasopressor use may be limited depending on institutions. Red Bull has been shown to increase BP in several studies. We aimed to evaluate the effect of Red Bull on perfusion-related variables after microsurgical breast reconstruction.

METHODS

We conducted a multicenter, prospective, randomized controlled trial. Female patients undergoing unilateral microsurgical breast reconstruction from June 2020 to October 2022 were randomly assigned to the intervention or control groups. The intervention group received 250 ml of Red Bull 2 h after surgery and twice on postoperative day (POD) 1. The control group received 250 ml still water at the respective intervals. BP was measured using a 24-hour monitoring device. Vasopressor use, fluid balance, and flap outcomes were compared.

RESULTS

One hundred patients were included in the study. Both groups were comparable concerning age, body mass index, and caffeine consumption. Mean arterial and diastolic BP were significantly higher in the Red Bull group after the second drink in the morning of POD1 (p-value = 0.03 and 0.03, respectively). Vasopressor use was similar, with a tendency for less postoperative etilefrine in the Red Bull group (p-value = 0.08). No flap loss was observed.

CONCLUSIONS

We observed increased mean arterial and diastolic BP in the Red Bull group after the second drink. Red Bull may be a useful adjunct after microsurgical breast reconstruction.

LEVEL OF EVIDENCE

I, therapeutic.

Comparing complications and perioperative teams in microsurgical breast reconstruction: retrospective cohort study

Background

Subspecialization with dedicated perioperative teams has become common practice in some surgical disciplines. While surgeon experience, the number of surgeons involved, and enhanced recovery after surgery (ERAS) pathways are known factors affecting the outcome after microsurgical breast reconstruction, the impact of the perioperative team has not been studied.

Methods

We conducted a retrospective cohort study consisting of a chart review of all patients who underwent microsurgical breast reconstruction from January 2019-April 2020. Surgery was performed by three microsurgeons at two institutions with different perioperative teams-one being a small clinic [private clinic (PC), 33 beds] and the other being a larger hospital [corporate hospital (CH), 335 beds]. Patients were grouped into two cohorts according to the institution where surgery was performed. The primary outcomes studied were frequency of revision surgery, flap loss and patient length-of-stay (LOS).

Results

One hundred and fifty microsurgical breast reconstructions were performed in 125 patients. Demographic data [age, body mass index (BMI), current tobacco use, donor site] was found statistically comparable between both cohorts. In the PC cohort with fewer perioperative care providers, lower rates of revision surgery and flap loss were observed (P value =0.009 and 0.04, respectively). LOS was not significantly different between the two cohorts (P value =0.44).

Conclusions

The outcome of microsurgical breast reconstruction depends on multiple factors. In this study, fewer flap complications occurred at the small clinic. One reason among others might be the lower number of perioperative care providers involved and hence higher likelihood of sharing microsurgical cases, which facilitates routine and ensures less variability in care. The value of perioperative team subspecialization in microsurgical breast reconstruction needs to be assessed in prospective studies.