How personalized medicine became genetic, and racial: Werner Kalow and the formations of pharmacogenetics.

Physicians have long puzzled over a well-known phenomenon: different patients respond differently to the same treatment. Although many explanations exist, pharmacogenetics has now captured the medical imagination. While this might seem part of the broader interest in all things genetic, the early history of pharmacogenetics reveals the specific factors that contributed to the emergence of genetics within pharmacology. This paper examines the work of one pioneering pharmacologist, Werner Kalow, to trace the evolving intellectual formations of pharmacogenetics and, in particular, the focus on race. Working in the 1950s and 1960s, Kalow made three arguments to demonstrate the relevance of genetics to pharmacology, based on laboratory techniques, analogies to differences between other animal species, and appeals to the logic of natural selection. After contributing to the emergence of the field, Kalow maintained his advocacy for pharmacogenetics for four decades, collecting more evidence for its relevance, navigating controversies about race and science, and balancing genetics against other possible explanations of patient variability. Kalow's work demonstrates the deep roots of the genetic and racial preoccupations in pharmacology. Understanding this history can restore attention to other explanations of individuality in medical practice, something of increasing importance given the current interest in personalized medicine.


In 1962 Toronto pharmacologist Werner Kalow published
Pharmacogenetics, the first full length treatment given to a new science that explored how genetic traits influenced the safety and effectiveness of medications. 1 When he presented his work at the New York Academy of Sciences that October, the New York Times ran both an article and an editorial about his findings. 2 Kalow's book drew interest at the time because it offered an explanation for something that increasingly vexed physicians. Physicians had known and expected for millennia that different people responded differently to their ministrations.
Skilled physicians tailored both diagnosis and therapy to each patient. With the rise of laboratory medicine in the late nineteenth century, however, standardized diagnosis and treatment increasingly became the goal of medical thought and practice. Patients' bodies did not cooperate: idiosyncratic reactions often accompanied the introduction of new drugs. 3 Even as physicians achieved unprecedented pharmacological capabilities in the decades after World War II, their actual experience with treatment outcomes forced them to re-engage with the puzzle of patient variability. How could it be explained? How could these explanations be used to optimize the safety and efficacy of medicine?
In the fifty years since Kalow published his book, genetics has come to dominate the answers to these questions. For instance, when Francis Collins and Margaret Hamburg, the heads of the National Institutes of Health and the Food and Drug Administration, described their vision for "personalized medicine," their concerns were exclusively genetic. 4 It is tempting to see pharmacogenetics simply as one of many examples of the current popularity of all things genetic. 5 Its actual emergence, however, was the result of deliberate strategies to demonstrate the existence and relevance of genetic determinants of drug response. When physicians and researchers began to focus on the puzzle of individual variation in drug response in the late 1940s, they had recognized three basic modes of explanation: heredity, environment, and behavior. Kalow's work demonstrates how one influential researcher made the case that one of these factors --genetics -deserved special attention.
Kalow did not rely on traditional techniques of genetic analysis that traced traits across families in order to demonstrate their inheritance. Instead, he developed three different kinds of arguments. First, he used new techniques of pharmacology and enzyme biochemistry, especially spectrophotometry, to quantify pharmacological individuality in ways that made difference legible, and eventually genetic. Second, he drew analogies between drug reactions in humans and other animal species: pharmacologists' use of a menagerie of animal models provided an influential precedent for thinking about differences between individuals and between species. Third, he appealed to the logic of natural selection and wove pharmacologists' findings into evolutionary narratives that justified the existence and relevance of individual difference. These last two arguments, in turn, introduced race into pharmacogenetics and forced pharmacologists to engage with the politics of race and science, a move with consequences that remain relevant today.
Kalow was not the only person responsible for the emergence of pharmacogenetics in the 1950s, nor was he the only person to make these sorts of arguments. He did, however, play a central role in the consolidation of pharmacogenetics as a specific field. 6 Despite this, he has received no serious attention from historians of medicine or science. 7 Anthropologists and sociologists, who have studied pharmacogenomics, 8 have focused on developments since the 1990s. 9 This paper offers a partial solution to this gap. By recovering Kalow's history, it offers insight into the interests and strategies that influenced the emergence of pharmacogenetics. And since his career spanned six decades, from its origins in post-war Berlin until his death in 2008, it also reveals important aspects of the subsequent development of the field. Two issues are especially relevant.
First, Kalow's work over his long career shows how race emerged as a central concern at the origins of the field and has remained an irresistible attraction for pharmacologists. This persistent presence offers a warning to current researchers who dismiss concerns about race and medicine by arguing that racebased prescribing is simply a passing phase on the road to a fully personalized medicine. Second, the current preeminence of pharmacogenetics was not foreordained. Although Kalow

An Accidental Pharmacogeneticist
In 1947, amid the chaos of post-war Berlin, two patients died after receiving routine doses of procaine [Novocain], a topical anesthetic. Hans Herken, a professor of pharmacology at the University of Berlin, suspected that endemic malnutrition in the devastated city had left the patients with a deficiency of a crucial metabolic enzyme. But he had no proof. He assigned the problem to a young pharmacologist, Werner Kalow. 11 Kalow had never intended to become a pharmacologist. As a medical student in Germany and Austria in the late 1930s, he had dabbled in research --for instance, conducting bird surveys on the Baltic coast in the summer of 1937.
Drafted into the German Navy in 1938, he was allowed to complete his medical studies. He submitted his thesis, on the effects of adrenal extracts on blood pressure, in 1941. While interning at a naval hospital in South Holland in 1942, he inadvertently insulted a visiting German admiral. The admiral promptly assigned 8 rates depended on how tightly enzymes bound their substrates. This, in turn, depended on enzyme structure. 21 Offered a faculty position in the Department of Pharmacology at the University of Toronto in 1951, and eager to avoid being drafted a second time-this time into the United States Army for the Korean War --Kalow moved once more. He worked on a variety of projects at first. One project, conducted with the Toronto police, used a Harger Drunkometer to quantify alcohol levels in intoxicated drivers after accidents. 22 He also continued his more traditional work in pharmacology and used a spectrophotometer to study curare and procaine. 23 Kalow soon returned to his true passion, cholinesterase. During the war, a Toronto team had used a cumbersome but precise gasometric technique to study enzyme activity. When Kalow demonstrated his spectrophotometer, they were impressed by its speed and simplicity. But they wondered whether it would also work on the rare patients that they had found with low cholinesterase activity. 24 This had become a problem of growing clinical interest. Donald Gunn, a psychiatrist at the Ontario Hospital in New Toronto, had recently begun using succinylcholine as a paralyzing agent for electroconvulsive therapy. The drug's fleeting action was well suited for the quick procedure. Since Gunn's patients received multiple treatments over weeks or months, it was easy to study their reactions to varied doses of the drug (Figure 1). Gunn had already found two patients who experienced prolonged apnea. One poor man remained paralyzed, while awake, for 45 minutes. Gunn and Kalow began collaborative studies, taking 118 patients through ECT treatments with doses of succinylcholine ranging from 30 to 1000 mg. 25 Results showed that the problem was not low enzyme concentration, but low enzyme affinity for its substrates. Since affinity depended on enzyme structure, and enzyme structure depended on amino acid sequence, the difference must have been genetic. When he presented his findings at the Canadian Physiological Society in October 1954, it made front-page news. 26 Kalow and Gunn expanded their studies to include 443 students, 346 laborers, 1228 people tested for syphilis, and 2442 psychiatric patients. In one extreme case, a psychiatric patient experienced 157 minutes of complete paralysis from a dose that generally produced just five to ten minutes of paralysis; it took a full eight hours for the patient to recover respiratory function. 27 Kalow pursued large scale testing so that he could be confident in his results before publishing.
The delay cost him dearly: he got scooped. In July 1956 a London group published in Lancet a report on a family with low cholinesterase activity. 28 Kalow felt the sting: "Such little incidents can be damaging to a scientist who thereby loses priority in public view --even if not in his or her own mind. Lesson to a scientist: always publish as fast as you can." 29 Kalow drafted a brief account of his own work and published it as a letter in Lancet that September. 30  Burnet, and James Neel. 33 Kalow read widely and learned much that seemed relevant to pharmacology. 34 Genes controlled antibiotic resistance in bacteria, chemotherapy resistance in tumors, and pesticide resistance in insects. One family of dogs, like his first patients in Berlin, had died from routine doses of procaine. He read about two recently described pharmacologic variants in humans. Some patients developed hemolytic anemia after primaquine prophylaxis for malaria. Others suffered hepatitis or peripheral neuritis after tuberculosis treatment with isoniazid.
Kalow also found inspiration in the old insights of Archibald Garrod about human metabolic variation. 35 Garrod, who studied cases of familial alkaptonuria, wrote in 1902 that metabolic variations between humans should be no more surprising than the familiar anatomic variations. 36 Scientists working for the League of Nations grappled directly with this problem as they worked to standardize measures of the potency of hormones and realized that drug response varied from person to person. 37 The existence of genetic influences on drug response seemed obvious, appropriate, and relevant.

Kalow compiled his research and literature review into his 1962
Pharmacogenetics. His readers could find everything from a primer about the genetics of bacteria, insects, and mammals, to discussions of abnormal drug reactions caused by metabolic variations, his own work on serum cholinesterase, and why asparagus altered the aroma of urine in some people but not others.
Although he knew that these were obscure topics, he saw much promise in the field: "One may ask whether the field of investigation here called pharmacogenetics is the end of a road or the beginning of a new one. There are many signs which suggest that the number of established instances of a hereditary control of pharmacologic responsiveness will considerably increase in the near future." 38 Thoroughness again cost him priority. As he worked on this long project, American geneticist Arno Motulsky published a 1957 article in the Journal of the American Medical Association that described the intersection of pharmacology and genetics, suggesting that adverse drug reactions could arise from "otherwise innocuous genetic traits or enzyme deficiencies." 39 German pharmacologist Friedrich Vogel coined the term "pharmacogenetics" in 1959. Despite this, Kalow's book still generated considerable interest. 41 When he presented his work at the New York Academy of Sciences in October 1962, it was picked up by the New York Times. As described in an editorial, "differences in genetic make-up between individuals make for radically different degrees of risk in taking drugs." 42 With the thalidomide scandals in 1961 and the Kefauver hearings from 1959 to 1962, there was great interest in anything that might improve drug safety.

The Technology of a Genetic Argument
It is worth looking closely at how Kalow made his case that individual variation in drug response was under genetic control, and how his strategies related to broader developments in the history of genetics and molecular biology. Kalow developed three arguments in parallel, based on the technology of biochemical enzymology, on an analogy to animal models, and on an appeal to the intuitive logic of natural selection.
The first argument depended on his laboratory practice. Kalow collected a blood sample and spun it down to isolate the serum. He diluted the serum and added a substrate, benzoylcholine. Using a spectrophotometer, he shined a light across the sample and recorded the absorption spectrum in order to measure substrate concentration. The rate of change of concentration over time measured enzyme activity. This was one of many ways that pharmacologists employed spectrophotometers in their research in the 1950s. 43 Kalow, however, developed a 13 special trick. He took a second sample and added both the substrate and an inhibitor --dibucaine. By repeating the measurements, he could measure inhibited enzyme activity. He called the degree of inhibition caused by dibucaine, expressed as a percentage, the "Dibucaine Number" (DN). Calculated in this manner, the DN did not depend on the concentration of the enzyme in the serum.
Instead, it depended on something more fundamental, the affinity of the enzyme's binding sites for different substrates. Binding affinity, in turn, depended on the physical structure of the enzyme. Since enzyme structure depended on its amino acid sequence, the structure and the resulting affinity and enzyme function had to be under genetic control. The DN was, as Kalow concluded, "genetically Laboratory at the University of London, described the value --to geneticists --of the identification of drug metabolizing variants: "Any gene may be abnormal in a particular environment. In this connexion, it is noteworthy that entirely unsuspected genetical sensitivities to artificial new environments produced by synthetic drugs have been brought to light." 55 Luca Cavalli-Sforza similarly described the interest of "genetical analysis" of drug response variants. 56 Kalow had a particular approach to genetics, one grounded in the concerns and techniques of medical pharmacology. His methods, however, allowed him to make powerful arguments about the genetic determinants of drug metabolism and engage the burgeoning community of molecular biology.

Animal Pharmacology and the Argument by Analogy
Even as Kalow explored the details of enzyme function and its genetic determinants, he also looked to the level of organisms and species in order to make claims about the relevance of genetics. When Kalow began his work in pharmacology in 1947, he quickly learned that different species responded differently to drugs and poisons. His initial study of the poison phenylethanolamine found markedly different effects in different animal species. 57 When he moved on to procaine, he learned that it too had species-specific effects.
Long used as a local anesthetic in humans, procaine had recently generated attention as a doping agent in horse racing: in horses, it acted as a central nervous system stimulant. 58 His work revealed a possible basis for this effect: human and equine procaine esterases had different affinities for different substrates. Humans, for instance, metabolized procaine 16-times faster than horses did. 59 This interest in species differences followed Kalow to North America. When he studied cardiac physiology in Philadelphia, his results differed from those reported by another team. He suspected that inter-species variation explained the discrepancy: "We have no explanation for the divergence of our findings (vasodilatation) from those of MacDowall (vasoconstriction) other than a species difference since his experiments were made on cats and ours on dogs." 60 When he moved to Toronto and began studies on curare, he found that its potency varied significantly between frogs and mammals. 61 During his initial years there, he had to earn extra money by testing horses at a local racetrack for morphine. Like procaine, it had stimulant effects in horses and had been banned as a doping agent. 62 Intrigued by these results, he surveyed the literature in search of other examples of differences in drug response between animals. He found many. The story began with rabbits. In 1852 a Viennese scientist noted that a population of rabbits survived on a diet of belladonna leaves. This was notable because belladonna, also known as deadly nightshade, contained atropine, an alkaloid that would be lethal to most mammals. Follow up studies in 1910 found that certain rabbits, but not all of them, possessed an atropine esterase that rendered belladonna harmless. Most scientists dismissed these findings. Kalow later speculated that "the time was not ripe for an acceptance of the idea that there are pharmacogenetic differences between various stocks of rabbits." 63 Subsequent study, however, confirmed the existence of the atropine esterase that allowed rabbits to thrive in fields of belladonna that would have left other animals wideeyed or dead. A 1943 study, for instance, found that roughly half of 181 rabbits tested possessed the enzyme. The prevalence varied between rabbit populations.
While "a New Zealand White race" lacked the esterase completely, other strains had it in many individuals, and still others, including "a pure chincilla race," produced both enzyme-possessing and non-enzyme-possessing bunnies. This pattern "suggests the existence of racial differences," at least among rabbits. 64 Kalow found many other relevant examples. Horses, humans, rabbits, and monkeys all had different levels of esterase activity; goats and cows had none. 65 Dogs were poor acetlyators. Sheep lacked glucose-6-dehydrogenase. Morphine response differed radically: "Some people always react to morphine like cats--with excitement; most people react like dogs and become sedated." 66 Species differences were most obvious with insecticides. The goal was to find chemicals that killed insects with minimal toxicity to humans and other mammals. Evidence of inter-species differences in drug metabolism posed a challenge for scientists who wanted to extrapolate from animal research to human use. As Kalow warned, "the fate of the drug may vary substantially from species to species, and conclusions reached by studying the fate of the drug in other species are not directly applicable to man." 67 Kalow also found evidence of variation at a finer level, between strains of the same species. Only certain rabbit races had atropine esterase. Harvard rats could tolerate a 44-mg dose of thiourea, but Hopkins rats could only tolerate 4 mg.
This was not a function of Harvard stalwartness: some wild-type Norway rats could survive 1340 mg. 68 When a laboratory catastrophe in Bar Harbor exposed countless mice to chloroform, mortality rates ranged between zero and 74% across different strains. Susceptibility to alcoholism varied as well. While many rats strains avoided alcohol, "one strain easily became drunkards when given a vitamin-deficient diet." 69 As Kalow later wrote, "The more precisely one measures the effects of a drug in laboratory animals, the more likely one is to find differences between strains." 70 To the extent that differences between (and within) animal species reflected genetic differences, then such species-specific differences in drug metabolism, especially among laboratory animals raised in controlled environmental conditions, must have reflected genetic differences.
Kalow was not alone in his interest in the precedent offer by species and subspecies variation in drug response. When Motulsky published his initial discussion of genetics and drug reactions in 1957, he admitted that no one had yet made a convincing case that genetic differences in drug metabolism existed in humans. However, he too knew about the story of rabbits and atropine esterase, and about strains of mice that were resistant to insulin. He described how "Qualitative differences between species also exist," with morphine, for instance, causing hyperexcitability in horses and cats, but not in humans. 71 Brodie had similar interests. In his introduction to the First International Pharmacological Meeting, held in Stockholm in 1961, Brodie noted that "Much work on the problem of species and strain differences remains to be done … Is it possible that enzymes which appear to be the same in two species are actually different?" Like Kalow, he had been struck in his own work by marked differences in drug toxicity between species: "I well remember our surprise when Dr. Axelrod and I found that phenacetin, an unusually non-toxic drug in man, was quite toxic in dog due to the accumulation of the methemoglobin-forming free amine in this species." 72 Although all were aware of species-level variation, it was Kalow who took the step of compiling an exhaustive review of existing knowledge of the differences in drug metabolism between species of bacteria, insects, and mammals, as well as between subspecies and strains of mammals. To the extent that metabolism, like anatomy, was under genetic control, the pervasiveness of metabolic variation demonstrated the relevance of pharmacogenetics.

Natural Selection and an Argument Based on Intuition
The ubiquity of drug metabolizing variants across the animal kingdom raised an obvious question: why was that the case? As Kalow recognized the prevalence of atypical esterases in 1954 and thereafter, he began to wonder about the meaning of the finding. Cholinesterase initially had no known function other than its impact on drug metabolism. Since it seemed impossible that drug metabolizing enzymes had evolved eons ago simply to be on hand once doctors developed pharmaceuticals, Kalow pondered what their physiological role had been in a world before procaine and succinylcholine. When his initial research suggested a higher incidence of atypical activity in psychiatric patients, Kalow suggested that cholinesterase played a role in neurophysiology and the pathophysiology of mental illness. 73 Subsequent research, however, did not bear this out. 74 His next hypothesis was more prosaic. Perhaps cholinesterase variants mediated response to potential environmental toxins, specifically some found in potatoes and asparagus. 75 Kalow's argument here paralleled one that was being made with increasing frequency in the 1950s and 1960s: that genetic variations existed because of evolutionary selective pressures on human and other animas. A.C. Allison had famously made the case in 1954 that sickle cell trait persisted in tropical populations because it provided selective advantage in settings of endemic malaria. 76 Kalow's examples, though less dramatic, followed the same logic. The enzymes that metabolized drugs must have evolved to manage various environmental toxins, especially those from food. Since different animals evolved in different environments, they evolved different toxin metabolizing repertoires.
When animals lived in the same environment, especially in controlled laboratory conditions, these differences became latent, only to be unmasked when exposed to drugs and other chemicals.
Some researchers were cautious when making these arguments. Montalenti had been puzzled by the way in which medical treatments had revealed unsuspected human variations: "it is noteworthy that entirely unsuspected genetical sensitivities to artificial new environments produced by synthetic drugs have been brought to light." 77 While he was not exactly sure what had maintained those alleles in the population, he suspected that the answer lay in natural selection: "it is reasonable to assume that any gene found persistently in human populations is likely to be in a fairly stable state with respect to selective forces." 78 Motulsky, citing the example of sickle cell and malaria, also suspected that "some selective advantage" maintained drug metabolizing variants in populations. 79 Others were more assertive. Brodie, whose work identified the liver enzymes responsible for drug metabolism, was more confident in his speculation: "One cannot help asking whether the latent capacity of these enzymes developed in the course of evolution to enable the organism to respond to the sustained presence of large amounts of foreign organic compounds ingested in food." 80 Drugs were simply examples "of a class of compounds to which the organism has been exposed over the ages." 81 Differential exposure over evolution had led to differential genetic capacity. Kalow's thought closely paralleled Brodie's. As Kalow later wrote: "Since drugs are part of the human environment, it is natural to encounter hereditary factors that affect the responsiveness of the human organism to drugs, and one should find these factors in all forms of life which are appropriately studied." 82 Kalow's interest in the evolutionary argument persisted throughout his career. For instance, in June 1997 evolutionary theorist Stephen Jay Gould published an extensive critique "ultra-Darwinian fundamentalists" who believed that natural selection was responsible for every complex trait. 83 Gould argued for a more pluralistic vision, in which natural selection was the most important of several mechanisms that generated new traits. This triggered an angry exchange with psychologist Steven Pinker about the merits of evolutionary psychology. 84 Kalow and a colleague entered the fray, offering pharmacogenetics as an example of the non-adaptationist mechanisms that Gould favored. A pharmacogenetic trait could emerge and persist, not because it conferred current selective advantage to an individual, but because it provided potential advantage to the species should the environment change: "they constitute a sort of biological insurance policy for the species, rather than for the individual. Like Professor Gould's 'spandrels,' they have no particular use when they arise, but may acquire a use later on." 85 Gould replied by tweaking their interpretation and suggesting that natural selection might favor "more variable species" (i.e., ones with more variable drug metabolism enzymes), a distinct mechanism from the usual focus on natural selection of individuals. 86

The Racialization of Pharmacogenetics
The analogy to species differences and the appeal to evolutionary logic both served to naturalize and justify genetic difference. They also naturalized racial difference: to the extent that different races had experienced different foods and environmental toxins, they too should have developed differently tuned drug metabolizing enzymes. This turn to race, at the very origins of pharmacogenetics, can be seen clearly in Kalow's work.
Kalow's interest in species and subspecies differences extended to studies of humans. As he surveyed the literature in the late 1950s, he found many reports about human racial differences in drug response. 87 H.A. Paskind, working in Chicago in the 1920s, showed that whites and blacks had different hemodynamic Motulsky was similarly aware of how drug metabolizing traits varied across ethnic groups. He began his 1957 discussion with the unusual prevalence of primaquine-induced hemolytic anemia in African Americans. He suspected that many of these ethnic differences reflected underlying genetic differences: "any drug reaction that is more frequently observed in a given racial group, when other environmental variables are equal, will usually have a genetic basis." 96 Even after the 1950 UNESCO statement on "The Race Question," that condemned the many myths and misuses of race, race remained an appealing analytic category for medical researchers. 97 Although Kalow knew about these varied examples, he had limited personal experience with racial variation. While stranded in the Japanese Empire during the war, he had encountered the "exotic" orient. His travel throughout Japan, China, Vietnam, and Korea, was "a truly broadening experience": "To us as newcomers, the first overwhelming impression was the strangeness of the country: the exotic faces, the traditional dress, the houses, the streets. It was all so different from home." 98 Toronto in the mid-1950s offered a more limited sample of research subjects. He tested people from Canada, Great Britain, Poland, Ukraine, Yugoslavia, Russia, France, Italy, Austria, Germany, Hungary, and "31 other countries (mostly European)." 99 When he found no differences between groups, he blamed the homogeneity of his research population: "The great majority of all investigated persons were of European extraction. There was no evidence for a different incidence of atypical esterase among Europeans, or between Europeans and North Americans, or between religious groups." 100 Even these limited studies, however, did occasionally yield a tantalizing lead: "We had an opportunity to observe atypical esterase in one Arabian family and one Persian." 101 Kalow's concern that the limited geographic diversity of his research subjects would limit their pharmacogenetic diversity reflects what was then a significant assumption about the distribution of human genetic diversity.
In his writings about these sorts of group differences, Kalow did not distinguish between species, subspecies, and races as categories of different ontological status. This was a question that scientists had debated since the eighteenth century. Beginning in the 1960s, biologists increasingly tried to enforce a distinction: species are a fundamental unit of biology, while subspecies and races are arbitrarily defined. 102 In this strict definition, the existence of species-level differences ought not provide a compelling analogy for the existence of subspecieslevel differences. Kalow, however, lumped the categories together, for instance offering explanations for "the different responses of breeds, strains, races, species." 103 Subspecies variations, such as those between white and chincilla rabbits or between Harvard and Hopkins rats provided a more credible precedent.
However, it remained unclear to what extent differences between inbred laboratory strains might parallel the natural variations between human populations.
Another point becomes relevant here. Scholars have described how model organisms --fruit flies, mice, viruses, and others --standardized biological research and facilitated the development of modern genetics. 104 Kalow's work demonstrates something different: the importance of lack of standardization. He and the researchers whom he read did not rely on one or a few animal models. Instead, they worked with a menagerie of different ones --mice, rats, cats, dogs, horses, and various strains of each. This diversity became relevant for Kalow's analyses.
Seeing heritable pharmacogenetic differences across so many different subspecies and species, Kalow expected that comparable racial and ethnic differences existed in humans.
Researchers' interest in race and pharmacology had real consequences. At the First International Pharmacological Meeting in 1961, where Kalow played an active role, Brodie posed the basic challenge: "An ever-recurring question is which species is closest to man in metabolizing drugs? One might indeed ask 'What man?', since individuals of homo sapiens metabolize a given drug at vastly different rates. In fact this biochemical variability is a major difficulty in evaluating drugs in man." 105 Kalow worried about this same problem: "gene frequencies vary between populations. This means that a drug that has been shown to be safe in one population is not necessarily that safe in another racial group." 106 Pharmacologists could not extrapolate between human populations, let alone between mammalian species, in their studies of drug safety and efficacy. When the New York Times first-year medical students to ingest a fixed dose of amobarbital; 129 of them submitted urine samples at 36 hours. 123 The researchers studied metabolite levels to determine relative activity of two specific metabolic pathways, C-hydroxylation and N-glucosidation. Most students exhibited twice as much hydroxylation as glucosidation ( Figure 4). Fourteen, however, had the opposite pattern. These atypical students caught Kalow's eye: "When preparing for the repeat investigations of students whose data seemed to be of particular interest, we were alerted by noting immediately and in succession some Chinese names among those with high N-glu excretion." 124 Four of the fourteen students with atypical activity were Chinese, as compared to only six among the 115 others, a difference that was statistically significant.
Kalow interpreted this finding with an odd caution. Even though he knew that "substantial racial or ethnic differences have been observed for the frequencies of all genes," he professed uncertainty about whether the racial variation in amobarbital metabolism was "unique or whether it represents a more or less common phenomenon." 125 Follow up studies confirmed the pattern. While 10 of 20 oriental students had this unusual pattern, only 1 of 20 Caucasian students did.
Kalow strengthened his conclusion: "Thus, there is no doubt about a racial difference in the mix of amobarbital metabolites." 126 As they continued to explore this question, Kalow's team used a specific graphing technique, a probit plot ( Figure 5), to transform overlapping Gaussian distributions (i.e., two overlapping bell curves) into distinct parallel lines. 127 This technique provided a starker visual argument for difference between groups. But once again he remained cautious: "The racial differences in the appearance of amobarbital metabolites represent an observation at the descriptive level without any definite assignment of causes.
While data collected among Caucasian twins and families suggested strongly that amobarbital elimination is under genetic control, such an observation made within one race does not prove a difference between races to be genetic. It therefore, should be emphasized that we cannot distinguish with certainty between environmental and genetic causes for the observations reported here. It nevertheless would be improper to dismiss the possibility of genetic interpretations." 128 While working on amobarbital, Kalow also used other drugs as metabolic probes. One, the antihypertensive debrisoquine, had been used by other researchers to reveal differences in a specific hydroxylation pathway between Britons and Kenyans. 129 He performed his own study of debrisoquine metabolism.
He studied 51 students, of whom 13 were "Oriental (Chinese) origin" and found that the pattern of metabolites was "influenced by the ethnic origin of the subjects." 130 This finding made sense: "ethnic differences definable by gene frequency occur in the cytochrome P-450 system as they do in all other, genetically controlled, drug-metabolizing enzyme systems." 131 Yet despite his growing confidence reporting in racial differences, Kalow remained defensive: "The data described here were not originally collected for the purpose of studying ethnic differences." 132 The ethnic difference "was discovered incidentally." Though wary about his small sample size, he reported the result because "the difference was statistically significant." 133 A third study, however, broke this pattern of racial findings. Other researchers had found differences in antipyrine metabolism between white Britons living in London and Indians living in India. Kalow recruited a sample of 48 Caucasian and 19 Oriental students and measured antipyrine urine metabolites at 48 hours. This time he found no differences between the groups: "The fundamental similarity in antipyrine metabolism in the two populations is remarkable, because metabolite patterns of antipyrine differed significantly between subjects, and because the same two populations are known to differ much in their capacity for other drug oxidations." 134 A parallel study, looking at antipyrine metabolism in rats, provided another surprise: "It must be noted that the metabolite profiles are remarkably similar in man and rat." 135 Kalow's writings in the late 1970s and early 1980s present a puzzle. He described surprise when he found racial variations in amobarbital metabolism, and surprise when he did not find them for antipyrine (and even more surprise with the lack of human-rat differences). When he reported his debrisoquine findings, he essentially apologized for doing so. These narratives of surprise and apology suggest that Kalow experienced discomfort about his pursuit of a palpably politicized science. 136 As he had first done in the 1950s and 1960s, he emphasized specific narratives that served to naturalize the findings of difference.
First, he worked with other researchers as they mapped ethnic and racial difference in drug metabolism on a global scale. They continued to find measurable differences whether they looked at the various cytochrome P-450 systems or at cholinesterase, paraoxonase, or acetylation. 137 Second, he fit these data into the old evolutionary stories: "It should not surprise us to discover that the optimal structure of our biochemical defence systems is somewhat different in people of different regions of the world. There are regional characteristics of flora and fauna, and hence there must be regional differences in naturally occurring toxic components of food or other environmental products." 138 Third, he preserved the slippage in his work between species and races. Commenting on the unusual extent of variation within human populations, Kalow noted that "Only a cosmopolitan species of brown bat seems to be genetically as diverse as are people within a given race." 139 Kalow officially retired in 1982 but remained active in the field. In 1985 he organized a major conference on racial and ethnic variations in drug metabolism that produced an influential collection of papers on the subject. 140 When the New England Journal of Medicine published a report on racial differences in propranolol metabolism, it recruited Kalow to write the accompanying editorial. He took the opportunity to reiterate his old argument: "For the physician, the lesson of the Zhou study should be the importance of increased awareness of the possibility of differences in drug response and in dose requirements among patients from various ethnic and racial groups." 141 He encouraged the World Health Organization and pharmaceutical companies to rise to this challenge: "The best scientific course for further action is to encourage systematic research to determine the factors, enzymes, and responses that tend to differ among the major races, and those that tend to be alike." As Kalow's work slowed, others took the lead and published ever more studies of racial differences in drug metabolism, 142 and then reviews based on these new findings. 143 Researchers continued to slip easily between studies on specific populations, such as Swedes and Han Chinese, to conclusions about races, whether Caucasian or Oriental. 144 Controversies about race even reached Kalow late in his career. By 2002 he had ceded lectures in the pharmacology course at the University of Toronto to his junior colleagues. Each year the course included a routine lecture on pharmacogenetics that covered ethnic differences in drug effects between Africans, Asians, and Europeans. In 2002, however, the students were outraged. As Kalow wrote, "The professor was shocked to experience ugly protests by some students who called him 'racist.'" 145 Kalow submitted an editorial to the local student newspaper, the Varsity, defending his colleague and the discussion of race: race was not a political issue, but simply a scientific one. He also tried to make the case more widely. Responding to a 2001 New England Journal of Medicine about the contested nature of race, he made his own case for why physicians ought to consider their patients' racial and ethnic backgrounds. The New England Journal rejected it without review. 146 He did succeed, however, in publishing a shortened defense of race in a specialty journal. 147

Genetics and Other Possibilities
Even though genetics in general, and race in particular, have received increasing attention as popular explanations for inter-individual variation in drug response, they have existed in parallel with other explanations. Recall that when Kalow was first assigned to the two cases of lethal reactions to procaine, his mentor, Hans Herken, had suggested that the problem was malnutrition. As Kalow pursued genetics, other researchers focused on nutrition and other environmental influences. In the early 1950s work on chemical carcinogenesis showed that specific drugs could influence the metabolism of other drugs, up-or downregulating drug metabolizing enzymes. 148 When the writings of Rachel Carson focused attention on environmental toxins, pharmacologists demonstrated how insecticides, including DDT, could induce the metabolism of drugs in humans. By the early 1970s many inducers of drug metabolism had been found, including insecticides, herbicides, oils, aromatic hydrocarbons, dyes, nicotine, preservatives, plant alkaloids, xanthines, flavones, ozone, and carbon monoxide. 149 It was this work, grounded in biochemical pharmacology, that identified liver microsomal enzymes, especially the cytochrome P450 system, as an important site for drug metabolism. 150 A New York group even demonstrated that food, whether charcoaled steaks, cabbage, Brussels sprouts, or the Atkins diet could accelerate the metabolism of widely used drugs. 151 These studies found high rates of intra-individual variability: as research subjects changed their diet, smoking habits, or other environmental exposures over weeks or months, their drug metabolism shifted as well. 152 Drug metabolizing capacity was not a fixed characteristics of individuals. Instead, it responded to individuals' chemical environments. Other studies found that apparent racial and ethnic differences, for instance between Londoners and Gambian or Sudanese villagers, also disappeared when diet and other environmental exposures were taken into consideration. 153 Kalow interacted with this environmental tradition throughout his career. In Cuban, Czechoslovakian, Danish, and so on, finishing with Vietnamese … It seems justified to assume that different people are routinely exposed to different enzyme inducers or toxic substances contained in their food and drinks. Thus, we explicitly recognize that different people are exposed to different environments, and that life in the same city does not indicate exposure to the same environment." 155 It made perfect sense to Kalow that the genetic and environmental traditions of drug metabolism research would co-exist. They were, after all, linked by the familiar evolutionary narrative. As animals evolved, they developed metabolic systems that could protect them from environmental chemicals and toxins. Natural selection favored metabolic diversity that allowed organisms to adapt to changing environmental conditions. 156 Kalow knew that the genetic and adaptive components could be difficult to sort out: "Many studies with probe drugs comparing different populations have given results that are equivocal in terms of the nature-nurture interplay." 157 He did not see this as a problem. Instead, it was simply an appropriately complex reflection of a complex system: "All these observations provide a cautionary note against simplistic genetic or non-genetic interpretations of ethnic differences in metabolism data." 158 From the 1950s into the 1990s, the research on the genetic and environmental determinants of drug response developed in parallel, often in synergy. A third discourse also emerged, one that that emphasized the importance of non-compliance. 159  inhibitors, antipsychotics, ß-blockers, calcium channel blockers, chemotherapy agents, diuretics, insulin, interferon, morphine, steroids, vasodilators, and warfarin. 163 The potential of commercial innovation based on such differences is best demonstrated by the emergence of BiDil, a combination of two generic medications that were repackaged and approved for treatment of heart failure only in patients who self identify as black. 164 Critics have argued that the advent of BiDil simultaneously represents both a cynical exploitation of loopholes in patent law and FDA policy, 165 and our "durable preoccupation" with race. 166 Despite the drug's commercial failure, there has been a "dramatic rise" since 2000 in biotechnology patents based on race and ethnicity. 167 Some products have already come to market. Companies now sell race-specific vitamins, cosmetics, ancestry tests, and even walking shoes based on the putative unusual thickness of American Indian feet. 168 The Scientists seek funding, prestige, and marketable products through genetics.
Regulators, investors, and patients encourage their pursuits.
Racial therapeutics have similarly strong appeals. Even as genetics has offered new modes of affinity, they have often become aligned with pre-existing ideas about race and ethnicity. 173 Race-based practice is even more convenient than genetic practice: doctors need simply to look at a patient and profile accordingly. 174 The resurgence of racial medicine has triggered understandable concern. Troy Duster, for instance, has warned that enthusiasm for genetics has led to the "molecular reinscription of race." 175 Duana Fullwiley has described this in detail. When the National Institutes of Health established its Pharmacogenetics Research Network in 2000, its stated goal was individualized medicine. This soon gave way to a pragmatic and value-laden acceptance of race as a proxy for individual genetic variation. 176 In response to the concerns, many advocates of pharmacogenetics and other forms of racial medicine reassure themselves by arguing that the use of race will be just a passing phase on the road to fully individualized treatment. 177 However, the long presence of race in pharmacogenetics and the enduring popularity of race as a concept in science and society more generally suggest that its appeals will not fade quickly. The continued presence of race in medical practice will require on-going scrutiny from medical and social scientists. 178 Even amid these many factors that facilitated the emergence of a racialized pharmacogenetics in the 1990s and 2000s, an appreciation of the longer history of the field deepens our understanding. Kalow