The 1942 Massive Contamination of Yellow Fever Vaccine: A Public Health Consequence of Scientific Arrogance

Abstract

In the late 1930s, the 17D vaccine against yellow fever was produced in record time. 17D was and is an excellent vaccine. Its rapid diffusion led, however, to several problems, the most important among them being the 1942 massive contamination of the vaccine distributed to the US Army by the hepatitis B virus. The US part of this story is relatively well-known, but its Brazilian part much less so. In 1940, scientists who were producing the 17D vaccine in Rio de Janeiro found that it was contaminated by an “icterus virus” that originated in normal human serum. They solved this problem through the exclusion of human serum from vaccine production, but failed to persuade their US colleagues to do the same. The Rio experts, aware of the potential pitfalls of a new technology, carefully supervised the consequences of their vaccination campaigns. They were thus able to rapidly spot problems and eliminate them. By contrast, US scientists, persuaded of their technical superiority and distrustful of warnings that originated from a “less developed” country, neglected to implement basic public health rules. A major disaster followed. (Am J Public Health. 2021;111(9): 1654–1660. https://doi.org/10.2105/AJPH.2021.306313)

Discussion of the manufacture of COVID-19 vaccines is dominated by praise for their rapid development and diffusion. This is, we are told, an entirely unprecedented feat. Such self-congratulatory statements are not entirely accurate. Earlier emergencies stimulated the accelerated development of some vaccines. The most recent example was the Ebola vaccine—rapidly developed, licensed, and tested in the aftermath of the 2013–2016 outbreak.¹

This essay describes an earlier event: the emergency production of the 17D yellow fever vaccine at the Rockefeller Institute virology laboratory in New York City during World War II. In 1941 and 1942, hundreds of thousands of doses of that vaccine were produced in record time and administered to US soldiers being sent to fight on the Pacific Front. Such rapid scaling-up of production of a viral vaccine was presented as an impressive endeavor. Alas, some batches of the vaccine were contaminated with a virus that induced jaundice (later identified as hepatitis B virus) that originated in one of the components of the vaccine, normal human serum. As a consequence, epidemiologists estimate that in spring 1942, between 40 000 and 50 000 US soldiers developed vaccine-related hepatitis.² The investigation that followed this outbreak acknowledged the existence of a severe public health problem, but presented it as unavoidable.³ In 1942, when the vaccine was mass produced at the Rockefeller Foundation, scientists were not aware of the existence of an “icterus virus,” transmissible by blood and other body fluids and resistant to heating to 56°C (then the usual method of inactivating pathogens in the serum), or of the presence of long-time carriers of such a virus. The investigation also praised the rapidity of the reaction of the Rockefeller Foundation virologists to this public health disaster. Merely a few weeks after the first report of jaundice in the US Army, they uncovered the culprit—human serum—and eliminated it from the vaccine’s chain of production.⁴

The story omits to tell, however, that the same problem had arisen a year and half earlier in Brazil.⁵ Brazilian and North American scientists who produced the 17D vaccine in Rio de Janeiro found that the most probable cause of jaundice linked with the yellow fever vaccine was contamination by an unknown virus present in the human serum, and that the manufacture of a serum-free 17D vaccine put an end to the incidence of postvaccination jaundice. The production of yellow fever vaccine in Rio de Janeiro was made in close collaboration with the virology department of the Rockefeller Foundation. Researchers working in Brazil repeatedly warned their US colleagues about the serious dangers of use of human serum. They were not believed: scientists in a cutting-edge biomedical research center were not inclined to take lessons from researchers from a “peripheral” institution. As long as the New York laboratory produced only a small volume of the 17D vaccine, the maintenance of serum in the vaccine’s composition probably had only limited consequences, but the situation changed dramatically with the mass production of this vaccine.

THE 17D VACCINE IN BRAZIL

The story of the yellow fever vaccine starts in the late 1920s, when successful infection of rhesus monkeys with yellow fever opened the way to production and testing of vaccines on the basis of killed or attenuated virus. The first vaccines made with a killed virus had limited efficacy only. In 1930, the virologist Max Theiler, who joined the Rockefeller Institute virology laboratory that year, successfully adapted the yellow fever virus to growth in a mouse brain, greatly facilitating its maintenance in the laboratory. Two years later, the director of the yellow fever laboratory at the Rockefeller Foundation, Wilbur Sawyer, together with his colleagues S. F. Kitchen and Wray Lloyd, developed the first live vaccine with an attenuated neurotropic yellow fever virus, 17E, grown in fertilized eggs. However, this virus was not sufficiently stable, and could be employed safely only when mixed with serum that contained antibodies against yellow fever.⁶

The virology laboratory of the Brazilian Yellow Fever Service at Oswaldo Cruz Institute, Rio de Janeiro (this laboratory was founded by experts from the Rockefeller Foundation), started a culture of the 17E strain in 1936. In 1937, a large epidemic of jungle yellow fever (yellow fever transmitted from jungle animals to people who live in proximity to a tropical forest) in the state of Parana led to a large-scale field trial of the 17E vaccine. Scientists who conducted this trial found that some of the vaccine’s recipients developed complications such as serum sickness, rash, and jaundice; these complications were attributed to reaction to anti–yellow fever serum.⁷ George Marshall Findlay and his colleagues from the Wellcome Foundation’s virology laboratory in London, who also noted jaundice among people vaccinated in Africa, similarly doubted the role of the yellow fever virus itself.⁸ Researchers therefore wished to develop a vaccine that could be administered without immune serum.

In 1937, Max Theiler, with his colleague Hugh Smith, developed another attenuated strain of yellow fever, 17D. Animal experiments and preliminary tests in humans indicated that 17D could be administrated without immune human serum.⁹ That same year, Smith traveled from New York to Rio de Janeiro, where he and a Brazilian colleague, Henrique de Azevedo Penna, perfected the culture of 17D in fertilized eggs, making possible rapid production of a new vaccine. The 17D vaccine was first used in Brazil during an outburst of sylvatic yellow fever in Minas Gerais in December 1937. In 1938, 1 058 328 Brazilians were vaccinated with 17D.¹⁰ Because of the unknown risks linked with distribution of a live vaccine for a dangerous disease, the Brazilian Yellow Fever Service staff elaborated a strict protocol of postvaccination surveillance. Each vaccination site had two tables: one for the preparation of the vaccine (diluted from a lyophilized stock) and another for records and paperwork. Data on each vaccinated person (name, age, sex, occupation, address) were recorded in a vaccination book. One copy of this book was kept locally and another in Rio. After a vaccination campaign in a given locality, physicians designated by the Yellow Fever Service visited this locality twice—once after a few weeks and then after a few months—to investigate the efficacy and potential secondary effects of the vaccine. The existence of detailed records of vaccination facilitated their task.¹¹

The protocol elaborated in Brazil for the early 17D vaccination campaigns was maintained during the impressive scaling-up of the vaccination. In 1939 alone, more than a million Brazilians received the 17D vaccine.¹² In 1939 to 1941, there were several incidents of iatrogenic effects of the vaccine. The existence of these incidents led a researcher of the Oswaldo Cruz Institute, Angelo Moreira de Costa Lima, to accuse the Rockefeller Foundation experts of using Brazilians as guinea pigs.¹³ This, however, was a minority opinion. The yellow fever vaccination campaigns were usually well accepted, probably because of fear of the disease, but also because the campaigns’ organizers successfully mobilized the support of local authorities and local elites. Moreover, although vaccination incidents were far from negligible, the existence of an efficient follow-up mechanism made possible rapid detection of a problem and its elimination. As the Rockefeller Foundation’s experts in Brazil explained, “the best protection against future accidents is the careful surveillance of vaccinated individuals.”¹⁴

The main pitfalls of using a live, neurotropic vaccine maintained through multiple passages in fertilized eggs, were that the virus might be too weak (excessive attenuation) or too strong (insufficient attenuation), and contamination of the vaccine by an external infectious agent. During the mass production of the 17D yellow fever vaccine in Rio de Janeiro, all three types of accidents did occur. In 1939, a vaccination campaign in the state of Espirito Santo was followed by an epidemic of yellow fever. The cause, researchers in the Rio laboratory found, was that the virus used for vaccination was weakened through too many passages in fertilized eggs. Accordingly, they limited the number of such passages.¹⁵ In June 1941, doctors observed a rise in incidence of encephalitis after a vaccination campaign in the state of Minas Gerais. An investigation concluded that the outbreak was induced by a batch of vaccine that contained an insufficiently attenuated virus. This incident led to a tightening of controls of attenuation of the virus through more frequents checks on its effects in laboratory animals.¹⁶

The most serious accident with the 17D vaccine in Brazil was, however, its contamination by a “jaundice agent.” During a vaccination campaign in the state of Espirito Santo in late 1939 and early 1940, a postvaccine follow-up identified more than 1000 cases of jaundice, with 22 deaths. The Rockefeller Foundation experts and their Brazilian colleagues decided to halt immunization with 17D until they could find the source of this jaundice. First, they excluded the possibility that the jaundice was produced by a mutation of the yellow fever virus itself. There was no correlation whatsoever between jaundice and the level of antibodies against yellow fever. They then carefully examined all the components of the production chain, and through an elimination process arrived at the conclusion that the culprit was normal human serum, employed as a suspension fluid that protected the fragile yellow fever virus.¹⁷ Virologists who had produced the yellow fever vaccine in Rio de Janeiro decided to discard all the old batches of vaccine. They imported a new strain of 17D from New York, and in the mid-1940s they started production of a vaccine in which the human serum was replaced by liquid from fertilized eggs. There were no more cases of vaccine-related jaundice in Brazil.¹⁸

The conclusion that serum can contain a jaundice-inducing “agent” was not very surprising. There had been numerous earlier reports of this phenomenon. Probably the most important among them was the description—made in 1919 by the South African professor of veterinary medicine Arnold Theiler—of jaundice in horses vaccinated against African horse sickness with a combination of live virus and horse immune serum. Theiler concluded that the jaundice—later named Theiler’s disease—was induced by a previously unknown virus in the horse immune serum.¹⁹ In 1939, Findlay and his colleagues proposed that jaundice observed in people vaccinated with 17D virus might have originated in normal human serum used to prepare this vaccine.²⁰ It is reasonable to assume that these studies were known to virologists confronted with the outbreak of postvaccine jaundice in Brazil. It is equally reasonable to assume that they were known to scientists who produced the 17D vaccine in New York, the more so because Max Theiler was Arnold Theiler’s son, and Findlay and his colleagues at the Wellcome Foundation had frequent exchanges with Rockefeller Foundation experts. Virologists from the New York laboratory were also aware of the decision, made in Rio de Janeiro in 1940, to produce a serum-free yellow fever vaccine. Nevertheless, a year later, when the New York laboratory began a massive production of yellow fever vaccine for the US Army, its vaccine contained normal human serum.

THE 17D VACCINE IN NEW YORK

From 1938 on, the Rockefeller Institute virology laboratory in New York was engaged in small-scale production of the 17D vaccine (Sawyer became the head of the International Division of the Rockefeller Foundation in 1935, but continued to work part-time in the laboratory). Production of the vaccine increased in 1939, but remained limited. Demand for the vaccine increased steeply in 1940, however, with the growing probability of the United States entering the war and opening a Pacific Front. Sawyer and his collaborators Johannes Bauer and George Strode were initially reluctant to scale up vaccine production in their laboratory. Faced with the alternative of either increasing their production of the vaccine or entering into partnership with private industry, they nevertheless chose the first alternative.²¹ In 1940, the virology laboratory’s researchers considered the possibility of switching to production of a serum-free vaccine, but finally decided that a period of rapid scaling-up of manufacture was not the right time to modify a product that, they were persuaded, was already satisfactory.²² Fred Soper, the head of the Rockefeller Foundation International Division in Latin America, strongly disagreed. The New York researchers believed that there were no complaints about their vaccine, he argued, because they did not make any effort to find out whether such complaints existed. In one documented case, a Pan American Airways pilot, M. Koepke, reported in 1941 severe icterus following vaccination against yellow fever, but the follow-up of this case was rapidly dropped.²³ In December 1941, Sawyer nevertheless reiterated that the absence of complaints about the New York vaccine proved that they were taking adequate precautions to avoid contaminations.²⁴

In 1941, the yellow fever laboratory of the Rockefeller Foundation was transformed into a production plant.²⁵ Between January 1, 1941, and April 9, 1942, it supplied the US Army with 141 batches of vaccine, that is, 7.7 million doses.²⁶ Whereas in 1941 the vaccine was mainly intended for the Navy and Air Forces, from December 1941, when the United States entered the war, the bulk of the vaccine went to the Land Army. The first cases of jaundice in the Army were reported in California in March 1942. Sawyer was at first reluctant to consider the possibility that the disease was induced by yellow fever vaccine. He and Bauer immediately traveled to California to investigate the jaundice outbreak, and on March 25 Sawyer wrote to Strode that he and Bauer were increasingly persuaded that it was a local event, unrelated to vaccination.²⁷ Things were, however, moving fast. In early April, jaundice cases among vaccinated soldiers appeared simultaneously in numerous sites. The majority of these cases were mild, but 17% were classified as moderately severe and 2% as very severe.²⁸ An emergency meeting of the New York virology laboratory decided on April 9 to halt the production of vaccine with human serum, and start the production of a serum-free vaccine.²⁹ On April 16, the surgeon general recommended stopping yellow fever vaccination until the arrival of new, serum-free batches.³⁰

On July 24, 1942, US Secretary of War Henry L. Stimson reported that from January to July, vaccination against yellow fever had induced 28 525 cases of “yellow jaundice” and 62 deaths. The Chicago Tribune of July 28, 1942, reacted to this statement, and to the announcement of the death from hepatitis of Lieutenant Colonel Edward Platt Reed, chief of the Inspection Division of the Chicago District. The Tribune editorial protested the Army’s decision to use a dangerous vaccine, adding that the majority of the vaccinated soldiers had a very remote chance of becoming infected with yellow fever. It concluded that the yellow fever vaccination was a sanitary disaster. Someone was guilty of a grievous error of judgment, and an inquiry was plainly needed to find who was responsible for this error.³¹ In his editorial of August 1, 1942, the editor of the Journal of the American Medical Association, Morris Fishbein, vigorously rejected the Chicago Tribune’s accusations.

There is every reason to believe that the occurrence of 62 deaths and some 28,000 cases of jaundice associated with the vaccination of millions of men is far less serious than would be an epidemic of virulent yellow fever among soldiers sent to tropical areas.

The Tribune editorial’s call for an inquiry, Fishbein wrote, “presupposes a stupidity on the part of medical science which is wholly unjustified.” By printing such accusations, the Tribune might create fear among soldiers, injure morale, and hamper the war effort.³² An article in the New York Times of October 18, 1942, stated that the “tremendous hullabaloo” raised by one Chicago, Illinois, newspaper about postvaccine jaundice was totally unwarranted. An investigation “has been unable to discover anything that savors of negligence or that gives any cause for alarm.”³³

There was no external inquiry into the postvaccine icterus incident. The report of an internal investigation by the Rockefeller Foundation, published in 1944, estimated that more than 26 000 soldiers developed postvaccine hepatitis.³⁴ Later, epidemiologists estimated that at least 40 000 soldiers were hospitalized with this diagnosis, of a total of 300 000 soldiers immunized with contaminated batches of the vaccine, the largest known iatrogenic incident of vaccination in US history.³⁵ The internal investigation conducted in the virology laboratory linked jaundice-inducing batches of 17D vaccine with serum from individuals who had had jaundice in the past.³⁶ In the second half of 1942, three surgeons from the US Public Health Service injected 278 volunteers from an “institution with a population of at least 1700” (probably a prison) with suspected batches of yellow fever vaccine or sera of people who developed jaundice following vaccination, and displayed a transmission of jaundice through serum.³⁷ The commission that, in 1944, investigated the massive contamination of yellow fever vaccine by hepatitis virus exonerated the Rockefeller Institute virology laboratory from responsibility for this incident.³⁸ Sawyer’s biographers acknowledge nevertheless that his involvement in the massive contamination of US soldiers harmed his reputation.³⁹ This also may have been the reason why he did not share the Nobel Prize awarded to Max Theiler in 1951 for his yellow fever studies.

The contamination of US soldiers by hepatitis B virus might have become a long-term health disaster as well. Because hundreds of thousands of US soldiers were vaccinated with contaminated vaccine in early 1941, there was a serious risk of increased incidence of chronic liver disease in this population. This probably did not happen. Investigations made 40 years later did not find a significant increase in severe liver disease among recipients of contaminated batches of vaccine; it was concluded that although a “natural” infection with hepatitis B through contaminated syringes, blood and blood product, and sexual relationships led to chronicity in 5% to 10% of cases, the proportion of chronic cases was much lower among contaminated soldiers, probably because they were young, healthy, and above all had a single exposure to the virus.⁴⁰ These studies nevertheless uncovered a small excess of deaths from liver cancer among the infected soldiers.⁴¹ One might add to those a possible excess of deaths from liver cirrhosis. In 2011, the Tampa Bay Times published an article entitled “St. Petersburg Woman Solves Mystery of Dad’s 1958 Death,” which reported that a woman whose father died from cirrhosis in 1958 heard about the contamination of soldiers with yellow fever vaccine, obtained evidence of vaccination of her father with a contaminated batch, presented this evidence to the Veterans Health Administration, and retroactively obtained veteran benefits for her mother.⁴²

CONCLUSION: TECHNOLOGY AND TRUST

Sawyer’s refusal to accept the Brazilian researchers’ view of the important risks linked to inclusion of human serum in the 17D vaccine was attributed mainly to his conviction that the Brazilian epidemiological data were not solid enough to justify a modification that might have decreased the potency of the yellow fever vaccine.⁴³ In a 1944 article, Sawyer and his colleagues explained that they knew that researchers in Brazil connected an earlier episode of postvaccine jaundice to the presence of virus-containing human serum in the vaccine’s production, but the “peculiar” traits of the Brazilian incident led them believe that in all probability “the yellow fever virus used in the preparation of the vaccine has become contaminated with an icterogenic agent during the cultivation in tissue cultures.”⁴⁴ Sawyer and his colleagues disregarded the Rio de Janeiro group’s claim that they carefully excluded the possibility of contamination of viral cultures themselves, probably because they did not trust virology studies conducted in Rio.⁴⁵ Accordingly, they became persuaded that the true cause of the end of the outbreak of postvaccine icterus in Brazil was not the exclusion of human serum from the vaccine, but the fact that the Rio group discarded their old viral cultures and started anew with a fresh strain of the virus imported from New York.⁴⁶

Sawyer and his colleagues attributed the problems with the vaccine produced in the Rio de Janeiro laboratory to failure to adequately supervise the virus cultures in an “inferior” setting. They were confident that such problems would not arise in the cutting-edge virology laboratory in New York. Confident of the high quality of their product, they did not believe that it was important to provide a careful follow-up of the people vaccinated. Their claim that they had never encountered a problem with the vaccine produced in New York was grounded in the absence of complaints, not in field studies. Scientists who worked in Brazil, the great majority of whom were locally trained Brazilian doctors and technicians, had a very different attitude. They employed low-tech approaches to quality control: careful registering of information on each vaccinated individual, systematic and thorough postvaccination follow-up, and meticulous epidemiological investigation of suspicious cases. Their main research tools were a “vaccination book,” a pen, and labor-intensive collection, transmission, and tabulation of data.

The yellow fever vaccine produced at Fiocruz (previously Oswaldo Cruz Institute) is still the 17D attenuated strain, and the vaccine department proudly preserves the first vials of the vaccine from 1937. In 2021, this vaccine is manufactured at the ultra-modern glass-and-steel building of Bio-Manguinhos, but this is a relatively new development. Until the early 21st century, yellow fever vaccine was produced in a modest building on the Fiocruz campus, lost among many other similar buildings, some dedicated to healing, some to teaching, and some to fundamental or applied research. The physical integration of the manufacture of vaccine with numerous other health-related activities was even more present in the late 1930s and early 1940s. Researchers from the Yellow Fever Service all took part in monitoring yellow fever outbreaks, producing and distributing vaccine, treating patients, and surveilling vaccinated people. The virology laboratory in New York, situated in the impressive building of the Rockefeller Institute—satirized in Sinclair Lewis’s novel Arrowsmith under the name McGurk Institute—was in the early 1940s at the cutting edge of biomedical science. It is perhaps not surprising that many (though to be fair, not all) of the leading scientists at that institution had little confidence in knowledge produced by virologists who were working in a modest laboratory in a peripheral country, relying mainly on elementary public health methods. Sawyer and other researchers at the Rockefeller Institute virology laboratory did not entirely disbelieve the results obtained in Rio de Janeiro; they just did not trust them sufficiently to act upon them and change their own behavior. In his Journal of the American Medical Association editorial of August 1942, Morris Fishbein criticized an assumption of “stupidity on the part of medical science.”⁴⁷ Stupidity is probably not the right term. Excessive faith in (presumably) superior technology, distrust of (presumably) less knowledgeable “others,” and neglect of basic public health approaches may be more accurate explanations.

Alas, doubts about the quality of studies made in the Global South—including those made in leading institutions by highly competent scientists—do not belong to the bygone past. In November 2015, two groups of Brazilian virologists—one from the Fiocruz Institute in Rio de Janeiro and led by Ana Maria Bispo, the other from Instituto Evandro Chagas in Ananindeua (Para state) and led by Pedro Vasconcelos—displayed links between the Zika virus and microcephaly. In Brazil, nobody doubted their findings; US experts were, however, skeptical. Peter Hotez, dean of the National School of Tropical Medicine at Baylor College of Medicine, explained in January 2017 that the initial reluctance of the Centers for Disease Control and Prevention (CDC) to accept the Brazilian scientists’ work slowed the international response:

even when the Brazilians found Zika virus in two women’s amniotic fluid and in the brain of a microcephalic fetus, the CDC would not accept it until they had done it themselves. I saw that as hubris.⁴⁸

The 2015–2017 Zika epidemic was mainly a regional phenomenon, with an especially high incidence in Brazil. The initial slowness of the international reaction to this epidemic therefore had a limited effect only on global public health. The slowness of European and North American reactions to Chinese publications from late January and early February 2020, which pointed to the danger of rapid propagation of a new coronavirus, had very different consequences.⁴⁹