AI to replace animals in testing — Dogs are Not for Testing.

This project was prepared and presented on 11 July 2018 to Ms. Harriet Green OBE, CEO of IBM Asia Pacific, by Shiranee Pereira, founder of Project D.O.N.T. Input received from Dr. Gargi Dasgupta, Senior Manager, IBM Research, Bangalore is gratefully acknowledged. It was taken up in 2019 by the ‘Science for Social Good’ group at IBM Research Laboratory, New York. A team of five research scientists headed by Dr. Payel Das worked on the project to bring out a brilliant machine learning model which can predict human toxicity with far more precision than animals. The findings were published in the paper ‘Accurate clinical toxicity prediction using multi-task deep neural nets and contrastive molecular explanations’.

The project seeks to augment the precision and prediction of drug and chemical toxicity by replacing the dog in testing with the power and potential of machine learning and artificial/augmented intelligence. The multi-billion dollar drug discovery industry — in which testing on animals is a mandated part, importantly the dog — if replaced by cognitive computing and Read Across strategies using Big Data, will have a robust, far more predictive, economical and humane replacement in computational toxicology. Test tried and validated by regulatory agencies, it will be an ingenious high-throughput model to predict toxicity with far more precision than animals.

Introduction

Imagine a world sans chocolate! That is exactly what it would be if we were to rely on testing chocolate on dogs. Chocolate can kill your dog — but chocolate for humans, besides being a relished treat, is good for the human brain. In dogs the ‘toxic’ component of chocolate is theobromine; while humans easily metabolize theobromine, dogs metabolize it slowly, allowing it to build up to toxic levels. Simply put, this is how unreliable and unpredictive the dog is when used in testing. Yet the dog remains the trusted yardstick and the last and final checkpoint to measure toxicity of a drug before clinical trials.

Archaic regulations — with no scientific rationale, promulgated more than half a century ago — still demand that every drug, pesticide, chemical and medical device is tested for safety on rodents and a non-rodent, most often the dog (otherwise the primate).

The cost of failure

Not surprisingly, today drug attrition hovers around 94–95%, taking over US$2.5bn and more than ten years to bring a new drug to market. Only one in ten drugs that enter Phase I of clinical trials reaches the patient. Nine out of ten of these drugs will be recalled at some point, as and when an unpredicted toxicity — not revealed by the animals the drug was tested on — harms, maims or kills a patient. The FDA determined that a ten percent improvement in compound attrition would save $100 million per drug. This staggering cost percolates to the end user, the patient, concealed in the cost of the drug, making essential and life-saving health care unaffordable to many, especially in third world countries.

This inability of dogs to predict human toxicity is mainly because of the significant differences between humans and dogs in their cytochrome P450 enzymes (CYPs) — the major enzymes involved in drug metabolism. An analysis of a comprehensive quantitative database of publicly available animal toxicity studies of 2,366 compounds indicated that dogs are highly inconsistent predictors of toxic responses in humans, and that their predictions are little better than tossing a coin. If a new drug has a 70% chance of not being toxic in humans, a negative test in dogs increases this probability to just 72% — essentially no additional confidence, but at great ethical and financial cost.

The ethical case

With more than 200,000 dogs being used annually, there has been public outcry opposing the use of ‘man’s best friend’ in toxicity testing. A campaign launched on Change.org called the Sound of Silence (S.O.S) campaign has 130,000 supporters to date. In recent years, neuroscientists using fMRI on fully awake and unrestrained dog ‘volunteers’ have fortified ethological findings on canine cognition and sentience — showing that dogs experience positive emotions and human bonding comparable with a human child aged 2–2.5 years.

The two round tables

Project D.O.N.T. was the result of two round table conferences held in India to discuss the use of dogs in testing in the light of recent scientific findings. Participants came from the pharma industry, testing laboratories, regulatory agencies and animal welfare groups. The first was held at the National Academy of Agricultural Research Management (ICAR) on 12 October 2017 at Hyderabad; the second at The Westin, Chennai on 17 May 2018. Nine of the twelve dog-using laboratories in India participated, alongside the Drug Controller General of India, the Indian Pharmacopeia Commission, CPCSEA, ICMR and IVRI.

Participants resolved that ‘the dog is a deeply sentient being’ and recognized an urgent need to chart a roadmap to replace the use of dogs in animal testing, and that they ‘see a promising future with computational toxicology as a stand alone or in combination with multiple-organ-on-chip as alternatives that could bridge the gap between rodent and clinical trials’.