Exploring the Power and Promise of In Silico Clinical Trials with Application in COVID-19 Infection

Background: COVID-19 pandemic has dramatically engulfed the world causing catastrophic damage to human society. Several therapeutic and vaccines have been suggested for the disease in the past months, with over 150 clinical trials currently running or under process. Nevertheless, these trials are extremely expensive and require a long time, which presents the need for alternative cost-effective methods to tackle this urgent requirement for validated therapeutics and vaccines. Bearing this in mind, here we assess the use of in silico clinical trials as a significant development in the field of clinical research, which holds the possibility to reduce the time and cost needed for clinical trials on COVID-19 and other diseases. Methods: Using the PubMed database, we analyzed six relevant scientific articles regarding the possible application of in silico clinical trials in testing the therapeutic and investigational methods of managing different diseases. Results: Successful use of in silico trials was observed in many of the reviewed evidence. Conclusion: In silico clinical trials can be used in refining clinical trials for COVID-19 infection.


Introduction
Computer simulations have been used in airplane and car design with great performance and accuracy for years [1,2]. Recently, companies and researchers have started to pay more attention to the use of computational modeling in areas like fluid dynamics, ventricular septal deceives, and movement mechanics of hip implants [3,4]. and submission of these studies [5].
Randomized clinical trials represent almost two-thirds of the approximately $2.6 billion needed to develop a new drug [6], it is a very long and expensive method for designing and validating new therapies and technologies with a very disturbing high failure rate [7]. Furthermore, real clinical trials may indicate a drug to be ineffective, however, it rarely indicates the reason behind this failure. This leads to the rejection of the drug despite the possibility of it working with small modifications [8]. Since then, the concept of changing clinical evidence with in silico evidence is slowly being more accepted, which was made evident by the FDA acceptance of in silico simulation as a possible replacement for animal studies in the assessment of artificial pancreas technologies [9,10].
In silico trials can be used for the noninvasive assessment of medical conditions. For example, in vivo computed tomography data were used to create individualized models that accurately predict the changes in vertebrae bone strength in mice [11], and the incidence of spine and hip fracture in humans [12]. From these observations, we contemplate that ISCTs can be used for choosing the specific mode of therapy with the appropriate doses for individual COVID-19 patients, depending on their individualized parameters like immunodeficiency and renal impairment. This will be a significant step in the path of individualized medicine. Furthermore, since this information can be available within hours to minutes to the clinical provider, this acts as a significant bonus when compared with the classical clinical trial.
This work aims to discuss the possible use of in silico trials in the design and testing of therapeutic and preventive measures to either reduce, refine, or partially replace human clinical trials done on the common disease with a focus on the possible applications of these methodologies in COVID-19 trials.

Types and phases of clinical trials
Clinical trials have been for a long time the golden tool for validating the efficiency of new drugs and biomedical products. It is a very long and complicated procedure containing five basic phases: 1. Preclinical phase: done in animals to understand the physiological action of the drug and four clinical trials in humans to assess the appropriate dose and duration of the drug and acquire data about short-and long-term side effects [13]. Figure 1 shows a schematic representation of the clinical trials pathway to further elucidate the relation between the different phases.
Although randomized clinical trials are the gold standard for evaluating the effectiveness and safety of the new medication and biomedical products, nonrandomized studies could still provide valuable and reliable input especially in fields such as forensic mental health where the randomized clinical trial may be inappropriate, therefore assisting in the decision for accepting or rejecting the new products or procedures [14].

COVID-19 conventional therapeutic clinical trials
In the past months since COVID-19 appeared from Wuhan, China, and spread around the world, over 3876 clinical studies have been recorded globally on clinical trial registries, including over 500 randomized controlled trials.  Such rapid development and inauguration of clinical trials are remarkable but presents challenges, including the possibility for replication and rivalry [15].
The most common therapeutic agent being trialed presently is hydroxychloroquine (261 trials with some trials planning to recruit over 25,000 participants), followed by antiinfective and antiparasitic agents as illustrated in Figure 2. and is largely effective, as verified by instances of repurposing treatments in cancer and other human diseases [16].
ISCTs could be very cost-effective in testing all possible therapeutic drugs for COVID-19 as it provides a fast and reliable method for acquiring data about the efficiency of these therapeutics with no or minimal human involvement.

COVID-19 vaccine clinical trials
The ideal design SARS-CoV-2 vaccine must address the need of vaccinating both the general population and high-risk individuals. Furthermore, the designed vaccines must adhere to the basic standard for vaccines like non-allergenicity, antigenicity, potency, and ability to induce both cellular and humoral response among others [17,18]. currently under preclinical evaluation with the majority (55 trials) using protein subunit platforms [19]. Figure 3 illustrates the distribution of these vaccines.

Artificial intelligence and COVID-19
Artificial intelligence (AI) has been useful in designing drugs and vaccines against several organisms, including bacteria and viruses that are known to cause severe infections. In particular, AI has a great outlook on developing vaccines against diseases such as HIV and malaria that have been problematic [20][21][22][23]. The SARS-Cov-2 is a single-stranded, enveloped RNA virus which has several antigenic proteins: the matrix (M) protein, nucleocapsid (N) protein, envelope (E) protein, and spike (S) surface glycoprotein [24]. The S protein has two subunits S1 and S2, majorly responsible for the viral fusion and binding, respectively.

Earlier applications
The process of developing new biomedical products requires three steps: design, preclinical assessment, and clinical assessment; all of which can be enriched by the use of in silico simulations [10].
In silico trials can be used to refine clinical trials and reduce the number and duration of animals and humans involved in this experimentation. For example, when a surrogate measurement is used in the in silco trial, the reproducibility of the in vivo studies is highly improved, and hence the numbers required for statistical significance will be reduced,

Future applications
Regular use of in silico software may become the common theme in medical research, where simulation could be provided for a large number of virtual patients ( <1000) in a matter of minutes to hours, with tools to validate and replicate these trials. Depending on the result of these in silico trials, new drugs and biomedical products can be approved by health agencies without the need for major human involvement in traditional randomized clinical trials. Table 1 summarizes the difference between ISCT and traditional clinical trials. Carlo simulations and R code to assess study designs and compare different dosing strategies using mixed-effects models. This will help in the optimization of the steps of the clinical trial and the prediction of the probability of success, the optimal dose, the cost-effectiveness, and finally to go or not to go decisions.
There are two available versions of the software -the basic free version and the expert version. While the basic version is suitable for small projects and demonstration purposes, the expert version provides more advanced tools like validation algorithms, which make it more acceptable for scientific publication (https://exprimo.com/simulo).
This software was used in 2018 in a study done by Murad Melhem and colleagues to model neutrophil response to granulocyte colony-stimulating factor (G-CSF) in patients with chemotherapy-induced neutropenia [28]. The successful use of this software in this work is encouraging and shows that it can be useful in other diseases like the current COVID-19 disease.

Highly Efficient Clinical Trials Simulator (HECT)
HECT is a web-based clinical trial simulator for the planning of adaptive trials (a type of trial that is more flexible than conventional clinical trials), written with statistical software

Previous studies on ISCTs
In total, six ISCTs were retrieved from the PubMed database. These trials were used for common diseases such as breast cancer, diabetes, and tuberculosis. The first trial simulates a clinical trial of immunotherapies in metastatic breast cancer, which was carried out by Hanwen Wang and colleagues. In this study, a quantitative systems Regarding the non-small cell lung cancer, an ISCT was taken, comparing the photon and proton radiotherapy effect on this patient group, which showed a reduction of the integral dose (ID) and the dose to the Organ at Risk (OAR) with protons therapy instead of photons even with dose escalation. In addition, simulation of clinical trials was conducted to identify and individualize optimal isoniazid doses in children with tuberculosis, where they concluded that in children, isoniazid should be optimized based upon disease process, age, and acetylation status.
Additionally, another ISCT was done, this time testing bioactive substance effect on healthy smokers. In this study, the result could explain the synergistic action mechanisms of the Sanghuang-Danshen (SD) bioactive in the regulation of vascular endothelial dilation, confirming the SD potential effect in releasing the vascular stiffness and decreasing blood pressure in healthy smokers. Finally, an ISCT with the University of Virginia tested the use of inhaled insulin using Type I Diabetes Simulator which in this study provides superior postprandial control and smaller risks of hypoglycemic events [30][31][32][33][34][35]. Table 2 summarizes these trials.

Results
Only six relevent ICTs were through PubMed. These trials were conducted for common diseases such as breast cancer, diabetes, and tuberculosis. The first trial simulateed a clinical trial of immunotherapies in metastatic breast cancer, which was carried out by Hanwen Wang and colleagues. In this study, a quantitative systems pharmacology model was built to integrate immune cancer cell interactions in patients with breast cancer, simulating central, peripheral, tumor-draining lymph node, and tumor compartments.
The proposed model provides a platform that can be further adapted to other types of immunotherapy which may contribute to the optimization of breast cancer treatment.
The other ICT was carried out by Susan G Hilsenbeck and C Kent Osborne in breast cancer, where they identified the role of adjuvant tamoxifen in progesterone-positive breast cancer. Table 2 summarizes the main aspects of the involved trials 4. Discussion

Validation of ISCTs
Validation of ISCTs is the procedure that evaluates the degree of how much the computer model and simulation agenda is capable of imitating a reality of interest.
It is a necessary step before application in clinical studies. This step can be done either through comparison with existing literature or through standardizing the methodology used in these trials. Assessing these two factors will help in proving the model integrity and hence the reliability of the ISCT [36,37].

Limitation of ISCTs
Resistance from the research professionals with limited interest in physics and mathematics, and the difficulty in simulating the complex physiological systems with the specialized advanced technical requirement to create new simulator software are the major hurdle of ISCT simulations.

Proposed ISCT protocol for COVID-19 vaccine
Building on the previous evidence, we believe ISCTs could be a plausible option in the current COVID-19 pandemic, as the software and models needed can be tested and validated against the preliminary results of the current COVID-19 clinical trials.
Specially designed models or an already available one like the model designed by Renz et al. simulating human alveolar macrophage with SARS-CoV-2 [38] can be used as a basis for the in silico trials. Furthermore, the Universal Immune System Simulator (UISS) platform which is an agent-based simulator is suggested by Giulia Russo and colleagues as a good choice for vaccine design studies [39]. Other platforms like Simulo and HECT can also be used in the current COVID-19 ISCTs with the following suggested parameters: 1. Type of trials: adaptive randomized ISCTs.
2. Number of patients: 10,000 virtual patients, between 6 and 59 years old who will receive two dosages of the vaccine or placebo.

Study limitation
The lack of standardized systemic review with appropriate statistical measurement in this work may undermine the acquired results and conclusion, however, the significance of this conclusion should not be overlooked.

Conclusion
ISCTs can truly transform the war against the COVID-19 pandemic. It is an attractive and applicable tool to accelerate the rate of approval for new therapeutics and vaccines for DOI  common and rare diseases. Furthermore, it is probably the most cost-effective method to reduce, refine, and partially replace COVID-19 conventional clinical trials ensuring both relevant results and minimal human risks.