Commentary: The hedgehog and the fox: Sir Harold Himsworth (1905-93)

The history of modern knowledge is concerned in no small degree with man’s attempt to escape from his previous concepts. Harold Himsworth1 Citation classics tend to be cited more often than read, and this has been the fate of the article by Harold Himsworth published in this issue of the IJE.2 Commentators (with what E.P. Thompson has called the ‘immense condescension of history’) are quick to award Himsworth retrospective marks for ‘getting it right’, but rarely pause to wonder what he was really trying to say, or why his ideas had so little traction for the best part of 40 years. Harold Himsworth was born to a Yorkshire family with no intellectual pretensions and his parents took him from school at the age of 16, ignoring his protests, to work at a worsted mill managed by a friend of the family. Resistance and undiagnosable illness followed until he was allowed to take a series of correspondence courses and enter University College Hospital (UCH) in London as a medical student in 1924. There he found that students were expected to learn everything about the preclinical sciences except why they might be interesting or relevant, a tradition that flourished at the same institution when I went there 50 years later. He kept such a low profile that the examiners needed to send for his photograph when he won the first of two University Gold Medals in 1929, following which he progressed to the Professorial Medical Unit at UCH (which had recently been transformed by a large grant from the Rockefeller Foundation) in 1931, taking over as Professor in 1939 at the age of 34.3 He spent most of the intervening time working on diabetes and the human response to insulin. The fox knows many things, but the hedgehog knows one big thing. Researchers can be classified in much the same way. Himsworth’s work on diabetes was characterized by dogged pursuit of a single goal, which classes him as a hedgehog. An insight into the nature of the man came in 1943. Workers in Glasgow had discovered that alloxan could produce diabetes in animals. The paper was written up and submitted to the Lancet, which sent it to Harold Himsworth for review. He considered it his duty to check the experiment for himself. It didn’t work, so he rejected the paper. When challenged by the authors, it emerged that he had accidentally inactivated the alloxan by boiling, and the paper was finally accepted.4 Diet and insulin were at that time the only treatments for diabetes. The clinical heterogeneity of the disorder was recognized, but this was assumed to be due to differing degrees of insulin deficiency. Investigators were puzzled by the fact that the islets looked relatively normal under the microscope with the staining techniques then available, and speculated as to a non-pancreatic cause for diabetes.5 Contemporary clinical research was based around clinical observation and the naked intellect, and Himsworth’s achievement stemmed from his willingness to stick to simple ideas, doggedly pursued through a long series of thoughtful investigations. Much of his experimental work was performed upon healthy males aged 18–22 (presumably medical students) ‘selected with special reference to the anatomical arrangement of the veins in their anti-cubital fossae and the shapes of the lobes of their ears’,6 (the ear-lobes required repeated puncture to obtain blood samples). His work was founded upon two major insights. The first was that the altered homeostasis of diabetes serves a functional purpose, and the second concerned the role of insulin action in achieving this functional equilibrium. These led to the first clear distinction between the two main diabetic phenotypes, and indirectly to the first use of the terms type 1 and type 2 diabetes. Years were to pass before the diabetes world reached similar conclusions by another route, and belatedly came to appreciate what he had achieved a generation earlier. His reason for opting to study diabetes came from the ward-based observation of a 30-year-old woman. He later commented that ‘I was struck by the fact that, although she would in a short time lose large quantities of sugar in the urine, this loss of sugar from the body did not reduce the blood-sugar level’. He tested this by giving her glucose, which was promptly dumped in the urine without producing a sustained rise in blood glucose. Next, he attempted to flush glucose out of her body by forced diuresis (based on an erroneous assumption, since you can increase fluid output but not glomerular filtration by doing so). Once again, circulating glucose levels were unaffected.7 He would certainly have been interested in more recent experience with the SGLT2 inhibitor drugs, which inhibit tubular reabsorption of glucose and are much more effective in increasing urinary glucose loss than lowering blood glucose, just as he would have predicted. The unmistakeable inference was that her level of glycaemia was under regulatory control. As he said, ‘It has been possible to trace the workings of an underlying purpose: that the head of pressure of sugar in the blood, in the circumstances existing at any particular time, is adjusted to maintain the adequate utilisation of glucose by the tissues’. The flux of glucose into cells, he considered, was maintained by the ratio between glucose and insulin. As insulin falls, glucose rises, thus restoring the ratio between [glucose] and [insulin] and sustaining the influx of glucose.8 The same principle was applied to the heart by another remarkable man, 20 years later. Cheated of a promising career as a cardiologist by a crippling attack of polio, Frank Guyton turned to physiology. Earlier physiologists believed that the heart controlled its own output, but he was to show that the heart actually has no say in the matter, and that the amount of work it must do is determined entirely by the oxygen demands of the cells. Himsworth believed that the liver and pancreas were similarly enslaved to the glucose requirements of the cells they served. The liver compensated for a failing pancreas by increasing its output of glucose. This implied that there was biological meaning in the apparently chaotic metabolic derangement of diabetes. Meaning meant mechanism, and mechanism meant therapeutic possibility. The concept of a metabolic set-point based around the balance between glucose and insulin was 40 years ahead of its time. Robert Turner reinvented the idea in 1976 and proposed that the equilibrium was based around insulin rather than glucose. On this interpretation, the body allowed glucose to rise in order to generate the insulin required for anabolic and metabolic purposes.9 Subsequent work clarified the hyperbolic relationship between glucose and insulin and forms the basis of our current understanding of the interaction between the two.10 Clinicians had long been aware of the distinction between young ketotic patients who died without insulin treatment, and older more obese individuals who kept going if they followed a diet. Elliott Joslin’s mother developed diabetes at the age of 60 in 1899 and survived the high fat/low carbohydrate diet he imposed upon her for 14 more years. She is described as Case #8 in his 1917 textbook. Such individuals were assumed to have partial insulin deficiency, based upon Frank Allen’s partial pancreatectomy model. Allen showed that if you removed around 85% of a dog’s pancreas, the animal could survive on a high-fat low-carbohydrate diet (the so-called ‘Esquimo’ diet), but would rapidly develop symptoms of diabetes if it ate normally.11 New-onset patients were accordingly admitted to the ward and fasted for a couple of days until the urine was clear of glucose. Carbohydrate was then reintroduced stepwise until sugar reappeared in the urine, and the amount of carbohydrate required would form the basis of their future diet. This was referred to as the glucose tolerance test. The same approach was continued, quite unnecessarily, following the introduction of insulin. By the late 1920s, brave investigators defied tradition and increased the amount of carbohydrate in the diet.12 As Himsworth remarked, ‘To everybody’s surprise it was found that in many diabetics this dietetic change necessitated no increase in their insulin requirements and in some cases actually led to a decrease’.13 He took the observation a step further by demonstrating that injected insulin produced a greater hypoglycaemic response in individuals treated with the high carbohydrate diet, thus demonstrating that diet could influence insulin sensitivity. The high carbohydrate diet worked because it allowed the flow of glucose to the tissues to be maintained at a lower head of pressure by making people more sensitive to their own insulin.14 The first practical application of his concept came as a new treatment for diabetic ketoacidosis. Insulin could snatch people back from the jaws of death, but many were lost through lack of expertise in using the new treatment. Himsworth saw that treatment should aim to feed the cells and not simply to lower the blood glucose. Give the cells glucose as well as insulin, and they will reduce their demands on the liver. He therefore developed a 10% glucose solution for administration with insulin in the treatment of diabetic ketoacidosis, forming the basis of a treatment regimen that we use in the recovery phase to this day.15 But why should the effect of insulin vary from time to time within the same individual? No one had as yet dreamed of an insulin receptor, and the tissues were therefore seen as purely passive targets of insulin action; there was no reason to believe that they might be able to adjust their response to insulin. The difference between insulin sensitive and insulin resistant individuals must therefore be due to factors influencing the activity of insulin when it reached the tissues, but not inherent in the tissues themselves. Himsworth accordingly proposed that insulin was released in an inactive form, and was converted into an active form when it reached the tissues. By analogy with the digestive enzymes described by Ernest Starling, Himsworth postulated the existence of an insulin activator produced by the liver, which he christened ‘insulin-kinase’. This idea, first proposed in 1933,6 haunted his thinking until he abandoned it in favour of another circulating antagonist, human growth hormone, later in the decade.13 Wrong though it was, the idea led to him to widen his search for insulin insensitivity. Many investigators had reported that their patients had widely differing insulin requirements, and Wilhelm Falta (1875–1950) proposed in 1931 that there were two types of diabetes, one sensitive to insulin and the other insensitive. This observation arose from the rigorous induction regimen his patients followed on the wards in Vienna. His practice was to feed them a constant diet until the amount of glucose in the urine became stable. Insulin was then introduced in small increments until glucose disappeared from the urine. The dose was now increased further until the patient had an episode of hypoglycaemia, a procedure referred to as the insulin tolerance test. The dose of insulin was then reduced by one notch and the patient was discharged from hospital. Sensitive patients needed only a few units of insulin to achieve this, whereas the insensitive ones took much longer to get into balance.17 Falta’s contribution to diabetes would be more widely recognized were it not for the fact that his Nazi party number indicated illegal membership prior to the Anschluss. His two sons were killed fighting for the SS during the war (personal communication, Michael Hubenstorf, Professor of Medical History, Vienna). Falta’s procedure was clearly far too cumbersome, and could not be used in non-diabetics. Himsworth therefore opted to give glucose and insulin at the same time—the glucose by mouth and the insulin by injection. The insulin dose was calculated as the amount needed to match the same glucose load in non-diabetics without producing hypoglycaemia. This simple test was the forerunner of the glucose clamp, the gold standard for measurement of insulin sensitivity, which delivers a constant intravenous infusion of glucose and measures the amount of insulin needed to hold the blood glucose at a given level. We are now ready to revisit Himsworth’s citation classic. In the casual manner of the time, this omits to mention how many people had been tested, makes no use of data or statistics and rests upon two ‘representative’ graphs. Insulin insensitivity having been demonstrated (he avoids ‘insulin resistance’ because the term had already been used in too many contexts to be helpful), he wanted to see whether tissue resistance to the action of insulin could be identified. Ingeniously, he tests this by comparing the arteriovenous difference in glucose across the forearm during the test. A wide difference appears in the insulin sensitive with little change in the insensitive, demonstrating that resistance to the action of insulin is present in forearm muscle. He goes on to suggest that the insulin insensitive type of diabetes might be the more common form of the condition, that it is due to lack of an insulin activator, and proposes that the sensitive form of diabetes will be more responsive to a high carbohydrate diet.2 In an important follow-up paper, he applies the test to healthy individuals ranging up to 65 years in age, and thus became the first to observe that insulin insensitivity declines with age, to the extent that some non-diabetic individuals show a similar loss of response to that seen in insulin insensitive diabetes, whereas many more fall into an intermediate range of insulin insensitivity which places them at risk of future diabetes.18 Here again he was 40 years ahead of his time. The dose of insulin required to balance the glucose load in non-diabetics proved equally effective in thin young people with diabetes, but was ineffective in older and fatter individuals. He inferred from this that the insulin sensitive group had diabetes because of lack of insulin, and that the remainder were resistant to its action. Not only did the test separate the groups, with almost no overlap, but it also corresponded broadly with clinical observation: The sensitive diabetics tend to be younger and thin and to have a normal blood pressure and normal arteries, and as a rule their disease is of sudden and severe onset. … The insensitive diabetics, on the other hand, tend to be older, obese, to have hypertension and frequently to exhibit pronounced arteriosclerosis, and in these patients the onset of the disease is insidious … Taken together, this evidence strongly suggests that two different types of diabetes exist, but sure proof that the two different responses to the insulin-glucose test distinguish two different disease processes can only be given by the demonstration of two different pathological mechanisms each giving rise to diabetes mellitus.19 Harry Himsworth left diabetes with a simple classic bedside experiment. It is still regularly cited for its clear demonstration that diabetes is not one disease but two. It was all but ignored by his contemporaries. In ancient Egypt they obliterated your existence from living memory by erasing your personal seal from the monuments. In modern science they do it by failing to cite your work. Elliott Joslin was one of the few who did quote Himsworth’s description of the two types of diabetes in the 1948 edition of his textbook, and mentions that his group had performed the ‘Himsworth test’ on 30 of their patients. The test did indeed divide them into insulin sensitive and insensitive patients, but ‘it became apparent that the results often cannot be correlated with clinical facts so that from a practical standpoint such tests are of no great assistance’. Himsworth had initially claimed that the test permitted a clear distinction between the two types of diabetes, but later conceded that ‘although there is a typical clinical group of cases giving each type of response, cases clinically atypical are frequently encountered in either group’.18 Joslin was probably not impressed for this very reason, and the study he describes was never published. The fallacy that a clear separation is possible between the two types of diabetes persists to this day.20 Forty-one years were to pass before Himsworth’s types were served up in a fresh guise and won immediate and universal acceptance. Why the delay? Why the acceptance? These are key questions, and the simple answer is that what we believe determines what we see. Joslin, like all other investigators of the time, was locked into the assumption that diabetes was an insulin deficiency disorder. Insulin resistance had to be rediscovered before Himsworth’s work could be appreciated. Himsworth already had a clear view of the two main types of diabetes by 1939, and his work gave birth to the terms type 1 and type 2 diabetes by an odd and serendipitous route.5 Constitutional medicine centred around the attempt to deduce disease susceptibility from phenotype. It dates back to Hippocrates but held powerful sway in the first half of the past century before evolving into anthropometry. W.H. Sheldon introduced somatotyping—the technique which gave us the terms ectomorph, mesomorph and endomorph—in 1940.21 The technique was performed by a trained observer, who scored the physique of subjects using a standard photograph measured against a grid. A New York physician called Draper invited a physical anthropologist to examine the patients attending his clinic in order to establish the somatotype of diabetes. The anthropologist found that that there were two somatotypes, not one. These he referred to as Group I and Group II.22 Some years later John Lister and other physicians at the Royal Free Hospital in London set out to characterize their own patients by combining the Himsworth sensitivity test with Sheldonian 1 and The Royal Free was a for and the did not how the working who their clinic were to for the technique a with insulin and a but more with insulin The commented that they had of group with normal blood pressure and an onset to the and the older, obese, group with hypertension and an insidious onset. … These types we have type I and type of the in paper. The show the standard for by of John Lister of glucose responses to the insulin sensitivity test. The insulin insensitive a increase in glucose with little or no response to insulin, whereas the glucose rise is almost in the insulin sensitive Many people have in the to Harry but special are due to John Himsworth and Michael of