When Animal Testing Failed

Case studies of drugs that passed animal safety tests but caused serious harm to humans—and how human-relevant methods could have predicted these disasters

92%
Drugs fail in human trials despite animal success
~2M
Hospitalizations/year from adverse drug reactions
$2.6B
Average cost per approved drug
50%
Of toxic drugs missed by animal tests

Major Drug Disasters

These drugs passed extensive animal testing yet caused catastrophic harm to humans

1957-1961
Thalidomide
Grünenthal (Germany)
In Animals

Passed safety tests in rodents with no signs of teratogenicity. Marketed as safe for pregnant women for morning sickness.

In Humans

Caused severe birth defects (phocomelia) in over 10,000 children worldwide. Limb malformations, organ damage, and deaths.

NAMs Could Have Detected

Human placental organoids and iPSC-derived embryoid bodies would show teratogenic effects on human cells.

10,000+ birth defects
2006
TGN1412
TeGenero (Germany)
In Animals

Safe in cynomolgus monkeys at 500x the human dose. No adverse effects observed in extensive primate studies.

In Humans

All 6 volunteers suffered catastrophic cytokine storm within hours. Multi-organ failure, ICU admission, permanent injuries.

NAMs Could Have Detected

Human immune cells in vitro would have shown the massive cytokine release response to this CD28 superagonist.

6/6 volunteers critically injured
1999-2004
Vioxx (Rofecoxib)
Merck
In Animals

No cardiovascular toxicity detected in preclinical animal studies. Appeared safe and effective for arthritis pain.

In Humans

Caused an estimated 60,000+ heart attacks and strokes before withdrawal. 3x increased cardiovascular risk.

NAMs Could Have Detected

Human heart-on-chip with vascular endothelium could model prothrombotic effects of COX-2 inhibition.

~60,000 heart attacks
1993
Fialuridine (FIAU)
Eli Lilly / NIH
In Animals

Safe in mice, rats, dogs, and monkeys for up to 3 months of treatment. No liver toxicity observed.

In Humans

5 of 15 patients died from liver failure. 2 required emergency liver transplants. Mitochondrial toxicity.

NAMs Could Have Detected

Human hepatocyte cultures and liver-on-chip show mitochondrial toxicity with human-specific transporters.

5 deaths, 2 transplants
1997-2000
Rezulin (Troglitazone)
Parke-Davis / Warner-Lambert
In Animals

Minimal liver toxicity signals in animal studies. Approved for type 2 diabetes treatment.

In Humans

Caused severe hepatotoxicity. 63 confirmed deaths from liver failure. Withdrawn from market in 2000.

NAMs Could Have Detected

Human liver organoids and hepatocyte chips detect idiosyncratic hepatotoxicity missed by animal models.

63+ deaths from liver failure
2016
BIA 10-2474
Bial (Portugal)
In Animals

Safe in mice, rats, dogs, and monkeys at much higher doses than planned for humans.

In Humans

1 volunteer died, 5 hospitalized with brain damage during Phase 1 trial. Irreversible neurological injury.

NAMs Could Have Detected

Human brain organoids and BBB-on-chip could reveal off-target effects on human neural tissue.

1 death, 5 brain injuries

How NAMs Could Prevent Future Disasters

Human-relevant testing methods detect toxicity that animal models miss

Human Liver-on-Chip

Detects hepatotoxicity with human CYP450 enzymes and bile transporters—catches drugs like FIAU and Rezulin.

Human Heart-on-Chip

Reveals cardiotoxicity and arrhythmia risk using human cardiomyocytes—would have flagged Vioxx.

Human Immune Cells

Tests cytokine release and immune activation in human cells—would have prevented TGN1412 disaster.

Brain Organoids

Models human neural tissue responses to drugs—could detect neurotoxicity like BIA 10-2474.

Placental Models

Tests drug transfer and teratogenicity in human placental tissue—would catch thalidomide-type effects.

Patient iPSC Models

Tests drugs on cells from diverse patient populations—reveals idiosyncratic and genetic-specific toxicity.