Rustam Gilfanov: What Do Automation and Big Data Offer to The Future of The Clinical Trials

First Posted: Apr 29, 2022 02:29 PM EDT
Close
Rustam Gilfanov

(Photo : Rustam Gilfanov)

A glance into the future: what might 2020s or 2030s hold for us?

Have you ever wondered how new medicines make their way into the market? Today pharmaceutical companies are required to hold volunteer-assisted clinical trials in order to prove their products are safe. Yet this is quite a recent requirement: some hundred years ago successful tests with lab animals were enough to start open sales. Which more often than it seems ended in tragedy.

Let's have a look into the history of clinical trials. Does it need to be changed taking into account modern advancements? In which way? Can big data influence the way clinical trials are currently held?

The pattern of clinical trials before present times

Lab animal tests cannot give any reliable picture as to the reaction of a human body to a certain drug. As obvious as it is now, this understanding needed years of trial and error to develop, and humanity had to pay with painful mistakes for it. 

More or less modern protocols of human trials were developed and applied in the 1950s. In 1954 the first extensive trial was conducted: a clinical trial of the polio vaccine. However, only in 1960s it became mandatory for an active ingredient of a new medicine to undergo clinical tests on human volunteers prior to mass production.

Back in 1937, S.E. Massengill Company produced liquid sulfonamide for curing infections in children. During development, they faced a problem: sulfonamide was difficult to dissolve. Developers chose diethylene glycol as the excipient and launched the drug into production. The medicine became widely available even without toxicity tests on animals, not to mention humans. After little time the Food and Drug Administration received a report about 8 children and one adult dead in Oklahoma after taking the drug. It was immediately withdrawn from the shelves, but not before it killed 107 people. 

After the thalidomide tragedy, it was not possible to ignore the situation anymore. In the lab, the tests of thalidomide on rats showed good results, so it quickly appeared in the pharmacies as a new sedative. With the support of West German authorities, the drug spread incredibly fast and was soon available in 50 countries. The tragedy did not make us wait long and 40,000 people were diagnosed with polyneuropathy, a disease affecting peripheral nerves. Some reports state that as many as 10,000 babies were born with serious limb deformities because their mothers took thalidomide. About half of the children died soon.

This tragedy pushed governments to take action and the development of medications was taken under control. The protocols for the clinical trial were developed and announced mandatory for all pharmaceutical companies. Later on, scientists learned that different animals could manifest different reactions to the same drug, so developers began to test substances on rats, hamsters, rabbits, monkeys, and other animals, as well as on cell cultures in vitro and on human volunteers.

Modern clinical trials: their types and methods

In the definition by the World Health Organization, we read that a clinical trial is a research study evaluating the effect of a medical intervention on the human health of an individual or a group of human volunteers.

Such research falls into two categories: observational and experimental. During observational trials, scientists gather data by observing a natural process and assuming that a substance under investigation helped faster recovery or led to negative effects. The experimental trial follows a different protocol. The volunteers are divided into two groups, one of which takes the drug being studied, and the other one gets a placebo or a different drug already commercially available. From here the procedures may vary depending on the type of drug and scientists can develop new approaches, discussing options with the academic community.

The document that describes different stages of tests is called a trial protocol. This is a mandatory document for the commencement of the trial. Among other information, the protocol describes the purpose of the trial, its location, the number of participants and how they were selected. All the volunteers for trials must understand they take part in a medical experiment and sign their consent for that. The principles of informed consent were developed by the World Medical Association Declaration of Helsinki. The signed and dated document must comply with a standard of GCP (ICH Harmonized Tripartite Guideline for GCP).

Besides that, the protocol must also outline procedures and methods of data processing and analysis. The modern plank of clinical studies is set at a randomized, double-blind, controlled clinical trial. It should consist of 3 obligatory and one optional phases. The first phase implies the use of a new substance on 15-200 healthy people, the second phase - on up to 300 patients with a condition the new medicine is supposed to treat. This could continue from several months to 2 years. The third phase is a comparison of the effectiveness of a new drug against existing standard treatment. It can include 300-3,000 people being observed for up to four years. The optional fourth phase can be held after the registration of a new drug. Its purpose is to expose adverse effects and contraindications, and thousands of people are usually taking part in it.

The final stage of a clinical trial is post-marketing monitoring. Collecting "field data" from people using the medication is a serious issue, but this is the most valuable information about the efficacy and safety of a drug. It is also a productive method to expand indications of a drug and reveal adverse effects.

As we can see, a comprehensive clinical trial for a drug can take up to 10 years before it becomes available to ordinary people. Before licensing a medication, authorities control that pharmaceutical companies follow the GCP standards. 

However, the procedure of the clinical trial can be sped up in case there is an urgent medical demand for a medication for a serious condition with few existing treatments. Often fast procedure applies to treatments for orphan diseases (conditions, diagnosed in 10 out of 100,000 people). Obviously, treatment of such rare diseases would not bring large benefits and not many pharmaceutical companies are willing to invest millions in the development of drugs for those. The faster process of licensing serves as a stimulus for the development of medications for orphan diseases.

Another example of fast authorization is the coronavirus pandemic. In December 2021, the European Medicines Association approved the conditional authorization of Nuvaxovid, a vaccine developed by the US company Novavax. Still, even though the drug becomes commercially available, the manufacturer must provide full information on its clinical trials. 

How to make medication development less time-consuming and expensive

In the end of 2010s several medications named "genetic pills" became available on the market. Among them Luxturna restoring eyesight, Glybera for treating lipoprotein lipase deficiency, and Kymriah fighting acute lymphoblastic leukemia. However, the price of these drugs reaches 1 mln dollars which makes them inaccessible to the majority of people. The price for Zolgensma, a treatment for spinal muscular atrophy in children, reaches incredible 2 mln dollars per dose.

To make these innovative treatments more accessible, we need to facilitate their faster exit to the market and reduce the amount of investments needed for licensing. Among other issues there are also a lack of volunteers, ethical and legal hindrances, inefficient time- and money- investments in the trial processes. Obviously, these issues result in enormous prices for revolutionary treatments.

How do we automatize the process using computer technologies?

Clinical trials can benefit from the application of computer technologies at different phases of the clinical trial. 

The start of the development of a medication is to distinguish the molecular target - a biological structure that needs to be affected in order to treat a certain condition. After the target is defined, the scientists start looking for a substance that can bind with it. The number of possible substances - organic compounds up to 30 atoms in size - is about 1063. With the use of a computer, it is not necessary to check them all. However, even applying in silico modelling in all projects (small molecules, proteins, or gene therapy drugs), scientists still need to synthesize and test hundreds of substances chosen by the computer. Out of those, only several make it to clinical trial. This time-consuming process asks for more automation.

With more automation not only we can analyze numerous substances in the same conditions much faster, but also eliminate the human factor. This requires modern computing methods and specifically machine learning.

Such deep embedding of automation would call for a unified laboratory system controlled by one software, which is presently almost impossible. Ideally, the lab system should comprise all the development stages from substance modelling to preclinical trials. Some companies have already been taking steps in this direction, like AstraZeneca with their NiCoLab solutions. In order to fulfil all the necessary tasks in the laboratory, it needs numerous sensors to control all the stages of experiments - a concept called the Internet of Things. A lab with such equipment is 30-40 times more efficient than a traditional one.

Even clinical trials themselves can benefit from automation: scientists can use real-time data from telemedicine devices on volunteers.

In silico: from animals to humans

Presently scientists have an opportunity to hold the preclinical trial in silico - that is to use a virtual model of a human organism for studying biological processes. Two other ways of study are in vivo (in molecular biology it is a colony of artificially cultivated cells) and in vitro (the system of cell-less synthesis in lab conditions). Within the in silico approach, we can model the behavior of one single molecule, a separate biochemical process, or even a specific physiological system. Developing these models is an expensive and complex task, however, the rewards are countless. In silico opens incredible possibilities for the research, testing of existing substances, and monitoring of various types of therapy.

The possibilities of big data

Clinical trials can be further accelerated with the help of big data. One of the most obvious use: modern analytical systems can choose the volunteers from several databases. As of now, the manufacturers of drugs are working to get access to the medical information of patients and starting partnerships with IT companies doing big data analysis. In February 2018, Roche pharmaceutical holding bought all shares of the Flatiron Health startup for 2 billion dollars. This is a company that gathers clinical data of people with cancer.

In South Korea two companies - Hanshin Medipia and Infinity Care - have already implemented Longenesis blockchain platform in their biomedical research. The platform deals with getting patients' consent for medical procedures within a study or a clinical trial. From the pharmaceutical company's side, the scientists have access to anonymous metadata and can offer the patient to take part in a study or ask for their feedback as to the drug use. Within the platform all the quotidian procedures are done much faster.

In some cases, big data gives an opportunity to forecast the results of the tests even without actually holding them, thus saving time and money. It is implemented by analyzing hundreds of substance's characteristics.

General data about the "field performance" of the drug is gathered during the fourth phase of the trial or after the start of sales. By 2017 already over 300 clinical trials implemented big data analysis (estimated by Reuters agency with reference to the website of global clinical trials clinicaltrials.gov). This research was mostly connected with cancers, heart diseases, and respiratory disorders.

In March 2018 several top pharmaceutical companies revealed they established specialized departments for gathering and using real-life data on various diseases. Specifically, it concerns the diabetes studies by AstraZeneca and Sanofi, joint research on stroke prevention by Pfizer and Bristol-Myers Squibb, and Takeda Pharmaceutical's study of intestine disorders.

Even though big data analysis opens huge possibilities, there are several restraints: quite often pharmaceutical companies cannot correctly analyze the data; the data itself can often be unstructured, incoherent, and presented in different formats, thus having no value; getting fully digitalized consent of patients for the use of this data is a big problem.

Personalized drugs 

Another branch of medicine with prospects for the use of big data is personalized medicine - the cutting edge of modern healthcare. Especially interesting is the sphere of development of drugs on demand. Even if it does not look obvious, big data can help tailor drugs according to the individual needs of people, offering maximum efficiency of the treatment. 

Using bid data analytics, scientists would trace patterns of disease development and predict possible mutations of viruses. On the other hand, it would be possible to calculate risks and, ideally, work on disease prevention or develop drugs that would be necessary in future. 

No one would argue that big data, automation, and digitalization would be a stepping stone for the development of modern medicine in its most fundamental aspects.

About the author

  Rustam Gilfanov, a successful IT entrepreneur, a founder of a prosperous IT company, and a partner of the LongeVC Fund.

See Now: NASA's Juno Spacecraft's Rendezvous With Jupiter's Mammoth Cyclone

©2017 ScienceWorldReport.com All rights reserved. Do not reproduce without permission. The window to the world of science news.
* This is a contributed article and this content does not necessarily represent the views of scienceworldreport.com

Join the Conversation

Real Time Analytics