← back to the index
RX-004 Synthetic estrogen 1971

DES (diethylstilbestrol) — the Useless Pregnancy Estrogen That Gave Daughters Vaginal Cancer

carcinogenicteratogenictransgenerationalendocrinologygynecologic-oncologyobstetrics1940s1950s1970s
Patients exposed
~5–10M U.S. exposed (mothers + offspring)
Documented harm
~1 in 1,000 DES daughters develop clear-cell adenocarcinoma (~40x baseline); broad reproductive harm across two generations
On market
1947→1971 indication (24 yrs); synthesized 1938
Status
Withdrawn

Summary

When the U.S. Food and Drug Administration issued its Drug Bulletin on 25 November 1971 declaring diethylstilbestrol contraindicated in pregnancy, it was retracting an indication it had blessed twenty-four years earlier for a drug that had never been shown to work — and the trigger was not a failure of efficacy but a cancer in a generation that had been exposed before birth. DES, the first orally active synthetic estrogen, was synthesized in 1938 in the Middlesex Hospital laboratory of Sir Edward Charles Dodds in London, deliberately left unpatented, and promoted from the mid-1940s by the Harvard husband-and-wife team of George and Olive Smith as a means of preventing miscarriage by "building up" placental hormones. The gap between that promise and the documented record is among the widest in pharmaceutical history: the drug did not prevent miscarriage, and it carried a harm that would not become visible for roughly two decades.

The signal arrived in April 1971, when Arthur Herbst, Howard Ulfelder, and David Poskanzer published a case-control study in The New England Journal of Medicine reporting eight cases of clear-cell adenocarcinoma of the vagina in young women aged 15 to 22 — a tumor so rare in that age group that a cluster of eight was itself an anomaly. Seven of the eight mothers had taken DES while pregnant; only one of thirty-two matched controls had. The association was overwhelming, with the odds of chance occurrence below one in one hundred thousand. The mechanism was transplacental: a drug given to the mother had reached across the placenta and altered the developing reproductive tract of the fetus, with the consequence emerging only at puberty.

The verdict is therefore plain at the outset. An estimated five to ten million Americans — pregnant women and the children born of those pregnancies — were exposed to DES between roughly 1940 and 1971 on the strength of an indication that controlled trials had already shown to be useless. As early as 1953, William Dieckmann's randomized, double-blind study at the University of Chicago Lying-In Hospital had demonstrated that DES did not reduce miscarriage, premature birth, or neonatal death; prescribing continued for eighteen more years regardless. The drug was not so much a wonder therapy that went wrong as an unproven one that was never stopped.

What remains is a two-generation injury with a long tail. DES daughters carry roughly forty times the baseline risk of clear-cell adenocarcinoma, develop it at a rate near 1 in 1,000, and face elevated rates of infertility, ectopic pregnancy, premature delivery, and breast cancer; DES sons and even a third generation have shown reproductive anomalies. The episode became the founding case study of the transplacental carcinogen and a standing rebuke to the idea that a drug's safety can be judged only in the patient who swallows it.

Timeline

1938
DES synthesized in London
Working in Sir Edward Charles Dodds's laboratory at Middlesex Hospital, chemists produce the first orally active synthetic estrogen; the discovery is published in Nature on 5 February 1938 and, by design, left unpatented, ensuring cheap, unbranded, ubiquitous manufacture.
1941
FDA approves DES for non-pregnancy uses
The agency clears diethylstilbestrol for menopausal symptoms, gonorrheal vaginitis, and other estrogen-deficiency conditions — its first U.S. approval, predating any pregnancy claim.
1947
Pregnancy indication approved
Following the Smiths' Harvard reports, the FDA permits DES for the prevention of miscarriage and pregnancy complications, opening the door to routine prescribing for "high-risk" and ultimately ordinary pregnancies.
1948–1953
Mass prescribing
The Smiths publish enthusiastic case series; DES is marketed under dozens of brand names and given to pregnant women across the United States, often at escalating doses.
1953-12
Dieckmann trial shows no benefit
William Dieckmann's randomized, double-blind study of ~1,600 pregnancies at the University of Chicago finds DES no better than placebo at preventing miscarriage or premature birth — the efficacy rationale collapses, yet prescribing continues.
1966–1969
A cancer cluster at MGH
Clinicians at Massachusetts General Hospital encounter an improbable run of clear-cell adenocarcinoma of the vagina in teenage girls, a tumor almost never seen before sexual maturity.
1971-04-22
Herbst study published in NEJM
Herbst, Ulfelder, and Poskanzer report eight cases; seven of eight mothers had taken DES in early pregnancy versus one of thirty-two controls, establishing the transplacental association.
1971-11-25
FDA contraindicates DES in pregnancy
An FDA Drug Bulletin declares DES contraindicated in pregnant women and notifies physicians nationwide; the indication that drove decades of exposure is revoked.
1971
Registry founded
Herbst establishes the Registry for Research on Hormonal Transplacental Carcinogenesis to track clear-cell adenocarcinoma cases in exposed daughters.
1978
European use persists
DES continues to be prescribed in pregnancy in parts of Europe until roughly 1978, extending exposure beyond the U.S. withdrawal.
1980-07-01
Sindell v. Abbott Laboratories
The California Supreme Court adopts "market-share liability," allowing DES daughters who cannot identify the specific manufacturer to sue makers in proportion to market share — a landmark in mass-tort doctrine.
1992
NCI Follow-up Study launched
The National Cancer Institute begins tracking more than 21,000 DES-exposed mothers, daughters, and sons to quantify the long-term, multi-decade burden of harm.

The Unpatented Miracle: How an Unproven Drug Became Routine

DES entered medicine as a triumph of chemistry and a vacancy of evidence. Natural estrogens were scarce, costly, and poorly absorbed by mouth; Dodds's 1938 synthesis produced a potent estrogen that was cheap, oral, and — because it was never patented — manufacturable by anyone. That openness, often praised as public-spirited, removed the single commercial gatekeeper who might later have policed its uses, and DES proliferated under dozens of brand names. The pregnancy application rested on a hypothesis rather than a finding: George and Olive Smith of Harvard proposed that miscarriage stemmed from inadequate placental hormones and that flooding the system with synthetic estrogen would sustain a pregnancy. Their published case series, lacking controls, reported encouraging outcomes, and on that foundation the FDA approved the pregnancy indication in 1947. The approval predated the 1962 Kefauver-Harris Amendments, which for the first time required manufacturers to prove efficacy, not merely safety — so DES reached pregnant women in an era when "does it work?" was not a question the law compelled anyone to answer. The marketing answered it anyway, in the affirmative, without the data to support it.

The Trial That Was Ignored: Dieckmann and Eighteen Lost Years

The case against DES as therapy was made not by its critics but by a rigorous test, and then disregarded. In 1953 William Dieckmann conducted at the University of Chicago Lying-In Hospital exactly the study the Smiths had never done: a randomized, double-blind, placebo-controlled trial of roughly 1,600 pregnancies. It found that DES did not lower the rate of miscarriage, prematurity, or neonatal death; some analyses suggested the drug arm fared slightly worse. By any modern standard the indication should have ended in 1953. Instead, the trial was absorbed as one voice in a debate rather than a verdict, the enthusiastic uncontrolled series retained their hold on prescribing habits, and DES continued to be given to pregnant women for eighteen more years. The cost of that inertia was not measured in heart attacks visible within months, as with later drug disasters, but in a latent biological change written into a fetus and dormant until adolescence. Every year the unproven indication survived added a cohort of exposed daughters whose risk would not declare itself until the late 1960s — the defining feature of a latent transplacental harm, where the interval between exposure and signal can swallow an entire era of prescribing.

The Reckoning: Eight Cancers, a Registry, and Market-Share Liability

The reckoning came from a tumor that should not have existed. Clear-cell adenocarcinoma of the vagina was, in young women, essentially unheard of; when Massachusetts General Hospital clinicians saw a cluster of cases in teenagers, the statistical improbability itself was the alarm. Herbst, Ulfelder, and Poskanzer's April 1971 NEJM paper converted anecdote into epidemiology: a case-control design showing that maternal DES use, taken before these patients were born, explained the cluster with a strength that ruled out chance. Within seven months the FDA contraindicated the drug in pregnancy. But withdrawal of an indication could not unmake the exposure already distributed across millions of pregnancies, so the reckoning shifted to surveillance and the courts. Herbst founded a registry to track cases; the NCI later built a follow-up cohort exceeding 21,000 people. In the courtroom, DES daughters faced a problem no prior plaintiffs had: with the drug sold generically under many labels decades earlier, most could not prove which company made the pills their mothers took. In Sindell v. Abbott Laboratories (1980), the California Supreme Court answered with "market-share liability," apportioning damages among manufacturers by their share of the DES market — a doctrine invented because the unpatented, unbranded design of the drug had erased the usual chain of proof.

Contributing Factors

01
Approval on an unproven hypothesis in a pre-efficacy era
DES received its pregnancy indication in 1947, fifteen years before U.S. law required proof of efficacy. The Smiths' theory that synthetic estrogen would prevent miscarriage rested on uncontrolled case series, and the regulatory regime of the day asked only whether a drug was safe as understood, not whether it worked. A claim was permitted to function as a fact, and millions were exposed on the strength of a mechanism that had never been tested.
02
A rigorous negative trial that prescribing ignored
Dieckmann's 1953 randomized controlled study showed DES did not prevent the outcomes it was sold to prevent. The persistence of prescribing for eighteen years afterward demonstrates that well-designed evidence does not stop a practice unless a mechanism exists to enforce it; absent withdrawal authority or mandatory re-review, a definitive trial became merely a contested opinion.
03
Latency that decoupled exposure from signal
The harm was transplacental and dormant: the drug altered a fetus, and the consequence — clear-cell adenocarcinoma — emerged fifteen to twenty years later at puberty. Standard pharmacovigilance, attuned to acute reactions in the patient who takes the drug, was structurally blind to an injury that skipped a generation and waited two decades. The latency itself was the concealment.
04
The unpatented, unbranded design that diffused both use and accountability
Left deliberately off-patent, DES was made by scores of manufacturers under many names, with no single owner to defend, refine, or restrict it. That same diffusion later prevented injured daughters from identifying who made their mothers' pills, forcing courts to invent market-share liability. A drug owned by everyone was, in practice, policed by no one.
05
A harm invisible to the consenting patient
The woman who swallowed DES was not the person who would be injured; the risk fell on offspring who could not consent and on a third generation beyond them. Safety frameworks built around the treated patient's own risk-benefit calculus had no place for an exposure whose victims were one and two generations downstream — the defining gap that the transplacental carcinogen exposed.

Aftermath

The material consequence of DES was an injury distributed across families rather than concentrated in a settlement. DES daughters carry roughly forty times the baseline risk of clear-cell adenocarcinoma, develop it at a rate near 1 in 1,000, and show markedly elevated rates of infertility, ectopic pregnancy, second-trimester miscarriage, premature delivery, and post-40 breast cancer; DES mothers face a modestly higher breast-cancer risk; DES sons show genital and reproductive anomalies, and studies have reported effects extending into a third generation. The durable ripple ran through both medicine and law. In medicine, DES became the textbook transplacental carcinogen and a permanent argument for treating drugs given in pregnancy as exposing the fetus, not merely the mother — a principle that reshaped teratology and prenatal pharmacology. In law, Sindell's market-share liability survives as a cited and debated solution to the problem of unidentifiable defendants in mass-exposure cases. The NCI's follow-up study, tracking more than 21,000 people since 1992, continues to publish, because the harm is still being counted decades after the last U.S. prescription. What remains is a byword: "DES daughters" names not only a cohort but a category of failure — the drug whose victims were born into the injury, whose efficacy was disproved before most of the exposure occurred, and whose verdict arrived a generation too late to recall.

Lessons

  1. Demand that an indication be earned by a controlled trial before mass exposure, not retired by one after it: if the evidence that would condemn a use could have been generated up front, treat its absence as a stop, not a gap to be filled later.
  2. When a rigorous negative trial lands, ask what mechanism exists to act on it — a study that merely joins a debate changes nothing; build the authority to withdraw or re-review so that "it does not work" can end a practice rather than prolong an argument.
  3. For any drug given in pregnancy, count the fetus and its descendants as exposed patients, and design surveillance long enough to catch harms that lie dormant for a decade or more; a vigilance window measured in months will miss a transplacental carcinogen entirely.
  4. Read latency as concealment, not safety: the absence of an early signal from a drug whose harm could plausibly skip a generation is uninformative, and you should weight biological plausibility above a reassuring short-term record.
  5. Trace where accountability lives before harm occurs: when a product is unbranded, generic, and made by many hands, identify who can be held responsible in advance, because a design that diffuses ownership will also diffuse liability onto the injured.

References