287

PREDIABETESA Synthesis

By W. P. U. JACKSON, M.A., M.D., M.R.C.P., D.C.H.Senior Lecturer, Department of Medicine, University of Cape Town, and Physician to Groote Schuur Hospital, Cape,

South Africa

II-

What It IsBefore you are born you are pre-natal, yet

already in existence; before you are diabetic youare prediabetic-a state which is not normality noryet a disease, but certainly there. This term' Prediabetes' thus connotes the state of a personduring the period before he or she becomes plainlyand clinically diabetic, in which, however, there isa latent abnormality which may show itself undercertain specific conditions. Since experts wouldnot agree on the precise definition of ' Diabetes,'or on what constitutes an abnormal glucosetolerance curve, it is clearly impossible to define'Prediabetes' more exactly. There are peoplewithout symptoms whose gross carbohydrateabnormality is discovered accidentally, but whoeveryone would agree are diabetic. Such peopleretain their carbohydrate disorder for the rest oftheir lives, and are not to be considered 'pre-diabetic '-their condition is better called mildasymptomatic' chemical' diabetes. The existenceof this phenomenon makes it impossible to knowin retrospect just how long a new patient has beentruly diabetic; in other words at what point herglucose tolerance became grossly impaired.In MothersWhen, with improved treatment, the pregnancies

of diabetic women came to term, it became plainthat the newborn were often dead, and/or exces-sively large. Skipper,1 25 years ago, was possiblythe first to observe that women may have bigbabies long before they developed overt diabetes.He and Priscilla White2 (1935) both suggested thatmothers who have large babies should be systema-tically examined for diabetes, but this was not doneon any scale until quite recently.

Several writers from different parts of the worldhave since confirmed the high frequency of over-weight babies (variously considered as over 9 lb.,io lb., 4or 4- kilos) and of late foetal or neonataldeath ('perinatal' loss).3-20There seems to be general agreement that the

perinatal loss in prediabetic years is around 15 to20 per cent., rising to as high as 30 to 50 per cent.in the two to five years immediately preceding thediagnosis of diabetes. A few workers, however,have found no evidence for an increased foetal lossbefore the appearance of diabetes. Thus Pirart21(in Brussels) observed no 'significant difference'from the general perinatal mortality in the pasthistories of 502 diabetic mothers, while Wilkerson22(in Boston) has found only ten perinatal deaths ini82 pregnancies in women who had previbuslybeen diagnosed as prediabetic (i.e. in a ' pro-spective' study). It is possible that the latter'sseries was unlikely to contain mothers who hadpreviously produced stillborn children; all thesame the explanation of these discrepant findingsis not clear.As far as I am aware, there is universal agree-

ment with regard to the tendency for large babiesto be produced in prediabetic years, and further-more that this tendency is plainly evident from thebeginning of the childbearing age, and shows butlittle increase with the approach of overt diabetes.In our own series,l? the incidence of babiesweighing over 10 lb. in the ten years immediatelyprior to the diagnosis of diabetes was 50 per cent.(by questioning), and was 35 per cent. in the period20 to 29 years preceding the diagnosis.Not all future diabetic mothers bear large babies

or stillbirths, even if they have many children.(We found that 62 per cent. of mothers in theirprediabetic years claimed to have produced at leastone baby over 10 lb. at birth.) Furthermore,there is no apparent relation between gigantismand viability of the child; in fact there is someevidence that the smaller babies (of diabeticmothers) are less viable.23 There does appear tobe some tendency for one large baby to be followedby subsequent large babies, and sometimes for thebirth weights to rise progressively, but these arenot strict rules.18 Thus it is quite common for ai2 pounder to be followed by a 7 pounder andlikewise for two or three stillbirths to be followed

by copyright. on 17 O

ctober 2018 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.35.403.287 on 1 M

ay 1959. Dow

nloaded from

288 .POSTGR.A.lJA'HTE MEDAL JOU.NA.L M.lay :959

MRS. F. AGED 60.

12 lb. t birth 'died ef diabee,,s' fii b 13gied 40 diabetic

L IVE CHILDRENSTILLBIRTHS. / LV CllRALL FULL TER / \ 0 ALL 11- 12 L

EIGHTS II -SLB miscrrige

+at le*at 4 other mircarriagee

0O Norial female Diabetic femaleQn Norlal male

Fic. .- Note that the proband's mother was diabetic, that she herself did not become diabetic untilfour years after her last child, and that even after so many stillbirths she can produce a stack oflive clhildren.

by a live child (Fig. I). in other words thismethod of analysis yields evidence of only oc-casional progression in the abnormal or damagingprediabetic factors, whatever they may be.

Other, more controversial obstetric associationsof prediabetes include infertility, miscarriage,toxaemia of pregnancy, hydramnios, prematurity,and congenital malformation in the foetus, andexcessive lactation.

Miscarriage. Since the evidence is still un-convincing that miscarriages are more frequentthan normal in established diabetics, it is to beexpected that it is also uncertain with regard toprediabetics. Pirart2l finds no evidence of anincrease. Nevertheless, on the assumption thatrepeated unexplained abortions might be a pre-diabetic or latent diabetic phenomenon, Williams,24Gilbert14 and Hoet25 have performed glucosetolerance tests in the mothers, and frequentlyfound them to be abnormal.

Toxaemia of pregnancy. Much the same posi-tion exists here as with repeated abortions-weare uncertain of the true relationship to diabetes orprediabetes. Again Pirart21 finds no increase.

Hydramnios. There can be no doubt thathydramnios is a very frequent concomitant of thediabetic pregnancy. Its incidence in prediabeteshas not yet been elucidated.

Prematurity. There is little if any tendency forthe diabetic's baby to be born prematurely;neither is the large infant of a diabetic post-mature26-it actually behaves more like apremature.

Congenital malformations. This is an importantsubject on which there is no unanimity of opinion.With r:egard to diabetics' c]hildren, several reports

have been published for and against the existenceof a high anomaly rate. P. White27 states thatmalformations 'have occurred in 80 per cent. ofthe infants of our diabetic mothers, compared withthe expected incidence of I.8 per cent. (sic.).The precise derivation of these figures is, however,rather uncertain. Recent careful analyses byCardell26 (on autopsy material) and Farquhar2(in live children and their families) have indicatedthat there is no excessive anomaly rate in'the offspring of diabetics. Hoet et al.28 claim to havefound a distinctly abnormal glucose tolerancecurve in 22 per cent. of the mothers of 50 childrenwith severe congenital malformation.

Diabetes occurring during pregnancy. It can nowbe categorically stated that glucose tolerance is notimpaired in a completely normal person duringpregnancy, or that any impairment is so slight asto be unimportant (Jackson,16 Hagen29 and un-published data). Consequently an apparently'temporary' diabetic state or significant impair-ment of sugar tolerance during pregnancy indicatesa state of potentially permanent diabetes in themother. In such a case the pregnancy hasbrought to light the prediabetic state. This isfrequently further indicated by the resulting foetusbeing large or stillborn. Sometimes a grossclinical diabetic state, even with ketosis, mayappear during gestation, yet the patient, and evenher tolerance curve, may be normal after parturition (Fig. 2).

Excessive maternal weight gain. It is common:place for women to start gaining weight excessivelyduring or soon after pregnancy. This tendencymay be exaggerated in those who later becomediabetic.3, 13, 30

by copyright. on 17 O

ctober 2018 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.35.403.287 on 1 M

ay 1959. Dow

nloaded from

May 1959 JACKSON: Prediabetes 289

M.L. AGE 22 .WEIGHT 11I LB.

SHOWING REMARKABLE DIABETOGENIC EFFECT OF' PREGNANCY

470:

440f

...........KT.L. .o,. .IN...... MONTS GNANT NOT PREGNANT.. 'NOT PN.EGNNT.CUR 'Now CURVE 'AGAIN ABNORMAL LCUREADAETIC)N...:O"NAL. FOLLO..wG

l*3o0 NOT PREGNANT. TS PEGAN (STILIRT AT CORTISONE.

ISMONTHI 'IN It 6Y MONTNS). (somg.z)PREGNANCY . 23 SVMPTOM+....- .--(CA.SARIAN OEL'VER.. E-LVELIVL LI 1Y)i 200

-I,

110

1956 NOV. 1956 MAY 1957 Y 19 OCT 1957 JAN. 1956SAD0 ZOE IS ABNOMAL * FASTING LEVEL * LEVEL 2 HOURS AFTER S9. GLUCOSE.

FIG. 2.-After this young woman had been in diabetic ketosis in late pregnancy, her carbohydratemetabolism appeared to revert completely to normal, though a subsequent pregnancy andthe use of the cortisone glucose tolerance test revealed the latent diabetic trait. (Glucosetolerance curves with half-hourly readings are shown.)

In FathersIt has been demonstrated16 21, 31 that diabetic

and prediabetic men father, on the whole, largerbabies than do non-diabetics. This larger size ofmen's babies is by no means as striking as in thecase of mothers (31 per cent. of babies of pre-diabetic mothers were over io lb. in my series,10.7 per cent. of (pre)diabetic fathers and 3.7 percent. of normal parents). On the other hand, thepresence of diabetes in the father had no ill effecton foetal survival.

In FamiliesDiabetes is certainly hereditary and generally

believed to be inherited through a common, auto-somal recessive gene. In view of the frequencyof the gene, a parent-child inheritance will oftenappear clinically (unlike the case in rarer recessivediseases). Roughly 20 per cent. of children withan affected parent may expect to become diabeticthemselves.27 Possibly there is more chance ofdeveloping diabetes if one has a diabetic motherthan a diabetic father.32 Prediabetes can thereforebe suspected in anyone with a strong family history,-and must actually be considered to be present incertain specific instances, as when an identicaltwin or both parents are diabetic.

In Patients with ObesityWhether obesity predisposes to diabetes, or

whether the prediabetic abnormality predisposesto the development of obesity (? through in-creasing appetite), or whether both things happen,is incompletely known. Certainly obesity isexcessively common in the diabetic clinic, milddiabetes is ameliorated or even clinically abolishedby reduction in weight,33 and several workers havefound a high rate of abnormal glucose tolerance inunselected obese patients, especially if they hadbeen obese for a long time (static obesity).3, 35A strain of obese mice become mildly diabeticunless their food intake in restricted36 while somepartially pancreatectomized animals may be madediabetic if they are overfed.37 On the whole itwould appear as if obesity, by increasing the totalbody mass and so the total metabolic turnover forwhich insulin is necessary, will expose those pan-creases whose reserve is insufficient to meet suchextra demands. After reduction in body weight,the pancreas may again become adequate for morenormal demands, and the ' diabetes' thus remits.

The 'Temporary Diabetes' Related to CertainOther Stresses

It is well known that there is a high incidence of

by copyright. on 17 O

ctober 2018 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.35.403.287 on 1 M

ay 1959. Dow

nloaded from

POSTGRADUATE MEDICAL JOURNAL

temporarily diminished sugar tolerance and evengross diabetes in patients suffering from acro-megaly, Cushing's syndrome, hyperthyroidism,severe fractures, severe bums, staphylococcal orother infections, undergoing treatment with gluco-corticoids or even after such a condition asmyocardial infarction. When investigations havebeen made, it has been found that there is a highincidence of diabetes in the families of persons soaffected.38 39 It seems likely that the particularcondition in each case simply renders overt alatent or pre-diabetic state.

What it does to the Foetus26, 40The diabetic's baby is generally abnormal in

several ways. It is overlarge, fat, rubicund andoedematous-looking (pitting oedema is not usual,however). Many of its organs are large, inparticular the liver, heart, and islets of Langerhans.The ovaries frequently contain luteal cysts andluteinised follicles.41 Active extramedullary haema-topoeisis is usual, even in the full term infant, andthe haematocrit reading is sometimes very high.The infant, even though large, behaves like a weakpremature; it respires with difficulty, suckspoorly, easily regurgitates fluid into its lungs and isliable to hyaline membrane disease. Hypogly-caemia is probably more extreme and longer lastingthan in other infants,23 but is now generallybelieved to cause no harm to the baby.We believe that most, and probably all of these

abnormalities may also occur in the foetus of theprediabetic. Most interesting to us have been theislets of Langerhans. Van Beek42 was the first toinsist that enlarged islets in a stillborn foetus meantdiabetes or potential diabetes in the mother.Woolf and I43 have confirmed this. We havefound the proportion of islet tissue in the pancreasof the prediabetic's foetus to be about 7 per cent.,as against about 1.3 per cent. in other foetuses.Furthermore the islets appeared to containa higher proportion of f cells than normal, and the, cells were unusually full of staining granules.This would seem to indicate a very high insulincontent of such a pancreas-even 30 times thenormal.The question of congenital anomalies is vexed,

as mentioned above. Hoet28 is most insistent thatthe prediabetic state accounts for many severecongenital anomalies (some 30 per cent. in fact)but this has not been confirmed.

Likewise no general agreement has been reachedregarding the usual state of the diabetic woman'splacenta. It is frequently stated to be heavier thannormal and to show advanced arteriosclerotic and'aging' changes and infarction.39,40 Recentinvestigators,286 53 however, have been unable tofind any significant difference from the placenta of

a normal mother. Consequently there are nogood pathological grounds for suggesting that anyfoetal abnormalities are produced by an abnormalplacenta.

Causation of the EmbryopathyThe first and obvious idea as to the cause of the

foetal abnormalities was that the maternal hyper-glycaemia allowed increased growth and perhaps insome other way damaged the developing foetus.However, the same sort of embryopathy occurs inthe prediabetic, in whom no hyperglycaemia canbe demonstrated at all. This theory must thenfall away.The next hypothesis concerned growth

hormone. Young4 and his co-workers had shownthat growth hormone could make dogs permanentlydiabetic, that it was in fact the so-called dia-betogenic hormone of the pituitary, and that itappeared to stimulate the islets of Langerhans tohypertrophy before the diabetes became severe,after which degenerative changes appeared in thep cells. The known pituitary hyperplasia inpregnancy, combined with the hyperplastic preg-nancy pancreas and the large size of the diabetic'sbaby, all seemed to fit in very nicely with the ideaof excessive growth hormone stimulation indiabetic pregnancy, Abaza, Varroud-Vial andRombauts46 actually described the ' Syndrome deYoung,' consisting of prolonged growth in women,high foetal mortality, large babies, hyperlactation,obesity and diabetes. Then Hultquist andEngfeldta7 reported the development of oversizedfoetuses in experimental animals given growthhormone during pregnancy, and the theory beganto seem plausible. Against it, however, was theabsence of acromegaloid features in the pregnantdiabetic, the infrequency of diabetes in humanacromegaly and-most damning-the absence ofparticularly heavy babies or foetal deaths in theacromegalic pregnancy.38 Furthermore growthhormone activity is associated with nitrogenretention rather than fat, and the muscles of thediabetic's baby are fat and flabby. Then no onehas been able to confirm Hultquist's work; in factsuch fine experimentalists as Jost48 have producedonly negative evidence. Until we can measurecirculating growth hormone, the question of itsparticipation in this syndrome is likely to remainin doubt.That corticotropin and the adrenal gluco-

corticoids can also act as diabetogenic agents thereis no doubt. Against the rather impressiverelationship of glucocorticoids to diabetes anddiabetic pregnancy are (i) the almpst invariablefailure to produce any similar embryopathyexperimentally, (2) the fact that babies born to

290 May I959by copyright.

on 17 October 2018 by guest. P

rotectedhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.35.403.287 on 1 May 1959. D

ownloaded from

JACKSON: Prediabetes

mothers with Cushing's syndrome are neither largenor dead (series collected by the author) and (3)the absence of any excessive rise in plasma cortisollevel during pregnancy in a diabetic.49, 50 In ourseries we actually found a lower mean maximumlevel of cortisol in diabetic and pre-diabeticpregnant women than in the non-diabetic group.At the moment we can only say that the partplayed by glucocorticoids in diabetic embryopathyis uncertain.What of insulin itself in this syndrome ? Is it

not tempting to think that those large islets playsome part ? Insulin is a growth hormone ;51 52 infact pituitary growth hormone will not functionwithout insulin. It seems possible that theabnormal size of the foetus might be due to foetalinsulin activity; but this would still leave unsolvedthe other features, and also the actual cause of theenlarged islets themselves.

Other suggestions have been made; e.g. anoxia,53abnormal placenta (discussed above). White27believes that a sex hormonal imbalance occurs indiabetic pregnancies, and she claims not only animproved live birth rate, but also a reduction infoetal size, by treatment with large doses ofoestrogens and progesterone. Obesity in themother may play a small part in some cases, sinceit has been shown that, on the whole, the largerthe mother, the larger the infant.16, 18 The role ofmaternal obesity, however, can be only a verysmall one. Heredity may also have some smallplace in the production of big babies, as indicatedby the studies on diabetic fathers (discussed above)and the work of Penrose54 on the inheritance oflarge-size infants in non-diabetic families.

Relationship between different parts of theembryopathy. Although the progeny of diabeticsand prediabetics are often large or dead, there isno close relationship between oversize and peri-natal mortality. Hoet claims further that thecongenital anomalies in the babies of the pre-diabetic occur quite independently of gigantism orperinatal mortality (except of course where theanomaly itself is lethal).

This makes one wonder whether the factorsresponsible for the gigantism are not quite differentfrom those which cause the stillbirth (or thecongenital anomalies).

Malins55 has observed an interesting relation-ship between the diabetogenic effect of pregnancyin the mother with established diabetes and thesurvival of the infant. He found that the foetuswas less likely to survive in those cases where themother needed a considerable increase in insulindosage during pregnancy, and particularly wherethis higher insulin dose had to be continuedafter parturition.

What it PortendsThe prediabetic state appears to extend back

indefinitely into a woman's obstetrical past,16 21, 5sits manifestations being limited only by the ageof commencement of childbearing. Thus womenwhose diabetes is discovered when they are 70, mayhave a history of large babies and even stillbirthsgoing back 30 or 40 years, although it is generallyagreed that the foetal loss is particularly high in thefive to ten years immediately preceding overtdiabetes. Evidence for the existence of the pre-diabetic state in early life can be obtained fromother quarters. Thus White27 has found analarmingly high proportion of abnormal glucosetolerance curves in young children of both diabeticmothers and diabetic fathers. It is plain that thegreat majority of these children will not becomeobvious diabetics until late in life (if they do so atall), yet they already manifest an underlying defectin their carbohydrate metabolism.From this and evidence discussed elsewhere38 it

may be postulated that the underlying abnormaldiabetic state exists from an early age (Fig. 3a),perhaps from birth, even when clinical diabetesdoes not appear until much later, and sometimesmaybe not at all.

Further it seems reasonable to considerthat diabetic vasculopathy is truly a part ofdiabetes;57 that a potential diabetic inherits notonly a congenitally defective pancreas but alsodefective blood vessels, particularly in certain sites,though it may equally well be that the sub-clinicalpancreatic defect has itself a deleterious effect onblood vessel walls, long before diabetes becomesmanifest. Certainly, then a poorly controlleddiabetic increasingly damages his already defectivevasculature, while careful control helps partiallyto protect it from further deterioration. If this istrue one could not in fact expect insulin or othermeans of' control' of diabetes entirely to preventvascular disease since the vessels are alreadydamaged by the time the diabetes is diagnosed.Is Pregnancy Diabetogenic ?Pregnancy (and other ' stressing' factors men-

tioned above) can plainly be considered as at leasttemporarily diabetogenic to the mother (Fig. 2).Thus it usually worsens hyperglycaemia andretinopathy in an established diabetic, it mayproduce retinopathy for the first time, which disap-pears after parturition,68 59 it causes hypertrophyin a normal woman's pancreas, it diminishesglucose tolerance in some apparently normalsubjects, and it may potentiate the diabetogeniceffect of cortisone or alloxan in laboratoryanimals.60 The question of importance is whetherthe diabetogenic action has a permanent effect--inother words whether pregnancy predisposes to the

May 1959 291by copyright.

on 17 October 2018 by guest. P

rotectedhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.35.403.287 on 1 May 1959. D

ownloaded from

2z2 POSTGRADUATE MEDICAL JOURNAL May I959

A CONCEPT OF DIABETES

OVERT

first pregnacy treated wth DIABETESilb stillborn cortisone boil

CLINICAL LEVELOF DIABETES

CLINICAL- RETINOPATHY

cnd

'SUBCLINICAL EDIABETES' NEPHROPATHY

BH 24 EARS 45 YEAR 55 EAS

-0

BIRTH 24 YEARS 4S YEARS 55 YEARSFIG. 3a.-' Iceberg' conception of diabetes: most of it is below the clinical level of overt mani-

festations, which here include embryopathy at the age of 24, transient hyperglycaemia duringstaphylococcal infection and while on cortisone therapy, and clinically apparent diabetesat age of 45, with retinopatry anc nephropathy at age o:f 55.

appearance, or earlier appearance, of clinicaldiabetes (Fig. 3b).

Suggesting that pregnancy is truly diabetogenicis the evidence that the incidence of diabetes ishigher in married women than single,61, 62 that itrises with increasing parity,63 and that it is veryhigh among women who have borne ten or morechildren." The latter author actually claimed thatover 50 per cent. of such women became diabeticin his series. If this can be confirmed it wouldindicate that multiple pregnancies are diabetogeniceven in the originally entirely normal (non-prediabetic) woman. However, another effectshould also be evident, namely the age of onset ofdiabetes should be lower with increasing parity.This has not been found.63Next we must enquire whether pregnancy is

diabetogenic for the child in the presumedabnormal internal environment of a diabetic orprediabetic mother. One certainly wonderswhether the grossly hypertrophied islets ofLangerhans in the infant should not suggest per-manent damage. If this is so then the childrenof diabetic mothers should have a greater tendencyto diabetes that the children of diabetic fathers.On the whole, however, it looks as if the abnormaldiabetic or prediabetic maternal environment mayproduce acute damage to the foetus, or even deathin utero, but that a surviving child usually recoverscompletely.Summary of Implications of Prediabetes

Diabetes is certainly hereditary (generally

considered largely to be a Mendelian recessivecharacteristic). From this alone, as Conn65 haspointed out, it might be stated that a diabeticpresenting at 60 must have had the diabetic'diathesis' all her life. Prediabetes, however,implies more than this. It suggests that some ofthe actual effects of the underlying condition arebeing produced before the clinical state is apparent,and that perhaps even the blood vessels are beingdamaged from an early age (Fig. 3a). I havesuggested38 that diabetes should be considered asstarting at birth (if not before), though we cannotdetect it till much later because of the crudenessof our yardstick of blood sugar level.

How to Diagnose itSuspicionThe discussion in the first part of this paper will

have indicated those people who may be suspectedof being prediabetic. The hints are:I. Positive family history, especially if close

relatives are affected, and especially if a childdevelops diabetes under the age of 6.66

2. Overlarge babies (especially if repeated).3. Perinatal mortality (especially if repeated).4. ? Repeated miscarriages; repeated pregnancy

toxaemia.5. ? Severe congenital anomalies in the child.6. Obesity, ? especially when developing rapidly

after pregnancy.7. ' Temporary diabetes ' during any acute stress.8. Glycosuria during pregnancy, which should

by copyright. on 17 O

ctober 2018 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.35.403.287 on 1 M

ay 1959. Dow

nloaded from

May I959 JACK(.SON: Prediabetes 393

IF PREGNANCY IS DIABETOGENIC

OVERT DIABETES

II lb. baby. stillborn.CLINICAL LEVEL

PREGNANCIES

BIRTH go 40 60 YEARS

EACH PREGNANCY BRINGS THE SUBCLINICAL PREDIABETES NEARER TO THECLINICAL LEVEL. SO THAT DIABETES WHICH WOULD HAVE APPEARED AT 60.OCCURS AT 45. ? DOES THIS HAPPEN.

FIG. 3b.-Should 3a really be drawn like this? If pregnancy is truly diabetogenic, each one willbring overt diabetes nearer, so altering the line of progression of the prediabetic state.

not be too readily shrugged aside as caused bya low renal threshold.

9. Xanthosis and mild carotinodermia.28io. Repeated attacks of acute pancreatitis.i i. Certain other conditions known to be fre-

quently associated with mild diabetes, e.g.cancer of body of uterus, gout.67

12. Renal glycosuria68 or spontaneous (nonin-sulinomatous) hypoglycaemia ('dysinsulinism'of Harris.69) 70

13. Typical diabetic-type vascular disorder (e.g.retinal aneurisms, or angina pectoris in apremenopausal woman).

Use of the Ordinary Glucose Tolerance TestSince 1952, the results of several series of oral

glucose tolerance tests in suspected prediabeticshave been reported.,6 19, 21, 22, 71, 72 Of course anumber of previously unsuspected, definite dia-betics have been discovered, whose sugar tolerancecurve no one would deny as being grossly ab-normal. Some were found without glycosuria,even during pregnancy, despite blood sugar levelsin the three hundreds. Such symptom-free,aglycosuric individuals might be considered'prediabetic' in a sense. Presumably theyshould be treated as established diabetics, thoughthe urine tests will be useless for control. I donot know of any follow-up studies in such cases.Most controversy has centred around those

tolerance curves which have been just outside theauthors' accepted normal limits. It has beenfound that the usual abnormality is a slightly highfigure 2 or 2i hours after the ingestion of glucose.In our own investigations we considered asabnormal a two-hour level above I40 mg. pero00 ml., with anything between I20 and 140 as

'suspicious' (capillary blood, 50 g. glucose, andmodified Hagedorn-Jenson glucose estimation).Occasionally the 2- to 2L-hour level was normal,while the fasting level was just over I20 mg., or thei-hour level above 200 mg. Such curves were alsoconsidered abnormal, but were much less common.Other workers, though using different techniques,have accepted very similar criteria of abnormality.

Series with adequate controls have demonstrateda statistically significant increase in abnormalcurves in women who gave rise to stillbirths(especially stillbirths with large islets of Langer-hans), to large babies, or who had glycosuriaduring pregnancy, or in men or women withdiabetes in their families. Despite this, and therather remarkable constancy of the abnormalitieson repetition of the test, many authorities at firstwould not believe that such minor changes intolerance as shown by 2- or 24-hour levels couldbe meaningful.73 The validity of such a belief,however, has since been vindicated by follow-upstudies which have demonstrated the appearanceof overt diabetes in many of those whose earlier

by copyright. on 17 O

ctober 2018 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.35.403.287 on 1 M

ay 1959. Dow

nloaded from

POSTGRADUATE MEDICAL JOURNAL

tolerance curves had been so slightlyabnormal.22, 65, 74, 75

It must not be supposed that abnormal curveswill be found in the majority of women who haveproduced large babies or stillbirths. For one thingprediabetes or diabetes is only one cause of thesephenomena, and secondly many women, eventhough prediabetic, may have completely normaltolerance curves, especially if they are tested whennot pregnant. It seems quite probable that theminor abnormalities in tolerance of' prediabetic'type are to be found chiefly within the five-yearperiod before overt diabetes appears, so thatpatients who are tested in an earlier prediabeticphase may show no abnormality in glucosetolerance.From what has been said above it is plainly

advantageous to perform repeated tolerance testsduring pregnancy, whenever possible. Abnor-malities are more likely to be found then, and aresignificant.The Cortisone (augmented) Glucose Tolerance Test

Berger (in I952)76 used corticotropin (ACTH)to ' sensitise' the glucose tolerance test, andFajans and Conn,7 in 1954, reported results usingcortisone for the same purpose. They investi-gated 152 healthy relatives of diabetic patients andfound 19 per cent. of these to be unknown dia-betics, as against i of 50 control subjects. Theythen tested 75 non-diabetic relatives of diabeticsby cortisone/glucose tolerance test and foundsignificant impairment in 24 per cent., as againstI of 37 normal controls. They further demon-strated that six patients whose carbohydratetolerance had apparently reverted to normal afterlosing weight nevertheless gave a positive responseto the test when this was augmented with cortisone.It appeared, therefore, that two distinct groups ofreactors to this test existed, and that a high pro-portion of positive reactors occurred in therelatives of diabetics. Were these people reallyprediabetics-the ones who would later themselvesbecome diabetic ? So far four out of 30 of suchabnormal reactors who have been followed up havebecome grossly diabetic.65

West,78 with similar but somewhat modifiedtechniques, confirmed the basic conclusion ofFajans and Conn. We have continued withFajan's technique the testing of certain specialgroups of patients (as yet unpublished data).Our tentative results so far obtained may benumbered:I. The mean rise in tolerance curve in apparently

normal people over 45 is higher than in thoseunder 45 (confirming West's finding).

2. The mean rise in normal pregnant women ishigher than in non-pregnant controls.

3. A proportion of mild diabetics show no rise atall after cortisone.

4. A high proportion of patients who are almostcertainly prediabetic (on obstetric groundsand/or on account of both parents or anidentical twin being diabetic) give negativeresponses.

It would therefore appear that this test, whileof some value in indicating the potential diabetic(Fig. 2), frequently fails to do so, while in olderpeople or during pregnancy it may appear falselypositive unless the levels considered to be ab-normal are raised.Duncan79 gives reasons for preferring to use

intravenous glucose tolerance tests, with thechange in increment index being the single-figurecriterion of abnormality. He found a positiveresponse in all mild diabetics and latent diabetics,and in 9 out of I9 suspected prediabetics. Thiswould appear to be a more sensitive method ofdetecting prediabetes.From the Pancreas of the StillbornAs discussed above, the islet hyperplasia in the

pancreas of stillborn infants is characteristic oferythroblastosis, diabetes or prediabetes in themother. If Rh incompatibility can be excludedthen it is extremely likely (? virtually certain) thatthe mother is diabetic or prediabetic. Woolf andJackson43 followed up 12 mothers whose stillborninfants had shown islet hyperplasia, without therebeing Rh incompatibility or maternal diabetes atthe time of the pregnancy. Eight of these 12 arenow plainly diabetic, and three others show ab-normal glucose tolerance curves of ' prediabetic'type. The final mother still has normal glucoseand cortisone/glucose tolerance curves, yet isalmost certainly prediabetic, since apart from herhighly suspicious obstetric history both her parentsare diabetic. It seems legitimate to conclude,therefore, that the finding of grossly enlargedislets in a stillborn is a very good way of diagnosingprediabetes in the mother, provided only thaterythroblastosis can be ruled out.A corollary of these conclusions is pointed out by

Jackson and Woolf,80 namely that from the pointof view of the morbid anatomist, a number of still-births for whom no adequate cause of death can befound, must be associated with prediabetes in themother. The findings of large islets on autopsy,then, would at least provide a partial explanation,and a new category, for these unexplained still-births.

For the assessment of the size of pancreaticislets (as a proportion of total pancreatic tissue)accurate methods are available,43 81, 82 whichindicate that the mean normal proportion is ai to 2 per cent. of total pancreas, while the

May r959294by copyright.

on 17 October 2018 by guest. P

rotectedhttp://pm

j.bmj.com

/P

ostgrad Med J: first published as 10.1136/pgm

j.35.403.287 on 1 May 1959. D

ownloaded from

May 1959 JACKSON: Prediabetes 295

' prediabetic' proportion is 4 to 15 per cent. Ingeneral, however, the enlargement of the islets isquite obvious on inspection, so that the term' continents' may be more suitably applied tothem.

Follow-upPlainly the final method of diagnosis of predia-

betes is to follow up the individual until shebecomes obviously diabetic. As indicated abovethis has now been done by several workers.

What Can We Do About It ?The study and diagnosis of prediabetes is not of

academic interest only. It is helping us to under-stand the meaning and natural history of diabetes,some aspects of its genetics, where best to look in asearch for new diabetics, factors in the developmentof the overt disease, the importance of so-called' temporary' diabetes, and the cause of certainstillbirths. From the point of view of the patients,can its diagnosis be of value in prophylaxis ?

Prophylaxis During Subsequent PregnanciesHaving seen the weak, oversized, Cushingoid

babies of some of our prediabetics, and bearing inmind their tendency to stillbirth and birth injuryand the improved results now being obtained bymodern management of the diabetic's parturition,I have no hesitation in recommending similarmanagement in the case ofthe prediabetic. In otherwords the pregnancy should be terminated aroundthe 36th to 37th week by surgical inductionprovided the foetus is believed to be of sufficientsize. In certain circumstances Caesarean sectionshould be resorted to. The newborn infantshould be treated precisely as though the motherwere a severe diabetic, with extra care in handling,oxygen tent, stomach aspiration, temperaturecontrol and no food for 48 hours. With thesesimple measures we have so far had completesuccess in prediabetics whose previous infants hadbeen stillborn, though our results are insufficientto be statistically valid (see also Carrington et al.20).Hoet goes further,28 and uses insulin during

pregnancy in his suspected prediabetics, eventhough there is no real hyperglycaemia. Hisdosage rises during the course of gestati'n to 6oor 80 units daily-according to the mother'stolerance. He claims to prevent miscarriages,stillbirths and congenital malformations.

Wilkerson's figures22 appear to show that insulin(2o units per day) during pregnancy in suspectedprediabetics will reduce the mean birthweight ofthe infants. In controlled series, 25 babies out of182 were over 9 lb. when the mother was untreated,as against 8 out of I62 in the maternally treatedgroup.

Prophylaxis against Future Diabetes, in Motherand ChildThe most obvious abnormality to treat in many

prediabetics is obesity. A diet of 800 calories aday of thereabouts is frequently necessary. Ifloss in weight is sufficient to prevent the onset ofovert diabetes it is not unreasonable to believe thatit also at least delays the vascular degeneration.The actual diet to be recommended to pre-

diabetics in general is controversial at the presenttime. Perhaps they, par excellence, should havea low saturated fat intake.Hoet and Wilkerson hope that insulin given

during pregnancy may also delay or prevent theonset of diabetes in the mother, and decrease thechance of diabetes occurring in the children. Thelatter possibility, of course, can apply only if thematernal prediabetic intrauterine environmentreally does predispose to diabetes in the infant.An obvious eugenic point might be mentioned

here. It would seem plainly inadvisable for twodiabetics to marry each other, or for a diabeticor prediabetic to marry into a family with severaldiabetic members. What with mixed diabeticclinics, camps for diabetics, the prolificity of themild and treated diabetic, and the improved foetalsalvage, the incidence of diabetes must surely beincreasing.Concluding RemarksWe are really only on the fringe of knowledge

and understanding of the fascinating state ofprediabetes. It is significant that the PublicHealth authorities (e.g. Wilkerson) have recognizedits importance and entered its field of research.If our appreciation of the condition seems frag-mentary at present, it must be realized that it isdifficult to grapple with a disease which is not there.

AcknowledgementsI should like to thank the innumerable people

who have discussed this subject with me, all ofwhom cannot be mentioned. Professor G. C.Linder, Dr. B. Lewis and Dr. N. Woolf have beenparticularly associated with me in our own work.I acknowledge the great help of Mrs. E. Orkin andDr. R. Hoffenberg with the manuscript.

Figures 2 and 3a are reproduced by permissionof the Editor of The Lancet.

REFERENCES-. SKIPPER, E. (1933), Quart. J. Med., 2, 353.- WHITE, P. (1935), Surg. Gynec. Obstet., 61, 324.3. ALLEN, E. (1939), Amer. J. Obstet. Gynec., 38, 98.4. MENGERT, W. F., and LAUGHLIIt, K. A. (1939), Surg.

Gynec. Obstet., 69, 615.. MILLER, H. C., HURWITZ, D., and KUDER, K. (i944),3.. Amer. Assoc., 124, 271.

S. MILLER, H. C. (I945), New Engl. .. Med., 233, 376.7. PALMER, L. D., and BARNES, R. H. (1945), West. J. Surg.,

53, 195.

by copyright. on 17 O

ctober 2018 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.35.403.287 on 1 M

ay 1959. Dow

nloaded from

296 POSTGRADUATE MEDICAL JOURNAL May i9598. HENLEY, W. E. (I947), N.Z. med. J., 46, 386.9. WHITE, P. (I947), Penn. med. J., 50, 705.io. KRISS, J. P., and FUTCHER, P. H. (I948), J. din. Endocr.S

8, 380.Ix. BARNS, H. H. F., and MORGAN, M. E. (I949), Brit. med. J.,

I, 5I.I2. PATTERSON, M., and BERNSTEIN, N. (1949), Arch.

intern. Med., 83, 390.13. GILBERT, J. A. L., and DUNLOP, D. M. (1949), Brit. med.

7., i, 48.14. GILBERT, J. A. L. (949), Ibid., i, 702.I5. MOSS, J. M., and MULHOLLAND, H. B. (i951), Ann.

intern. Med., 34, 678.I6. JACKSON, W. P. U. (1952), Brit. med. J., 2, 690.17. JACKSON, W. P. U. (I953), S. Afr. med. J., 27, 795.I8. PIRART, J. (1954), Ann. Endocr. (Paris), S1, 58.19. CARRINGTON, E. R., SHUMAN, C. R., and REARDON,

H. S. (1957), Obstet. Gynec., 9, 664.20. CARRINGTON, E. R., REARDON, H. S., and SHUMAN,

C. R. (x958), J. Amer. med. Ass., I66, 245.2r. PIRART, J. (x955), Acta. Endocr., 20, 192.22. WILKERSON, H. L. G., and REMEIN, Q. R. (1957), Diabetes,

6, 324, and personal communication.23. FARQUHAR, J. (I958), personal communication.24. WILLIAMS, E. C. P. (1933), Lancet, ii, 859.25. HOET, J. P., and BRASSEUR, L. (I954), Ann. Endocr. (Paris),

IS, 26.26. CARDELL, B. S. (I953), J. Obstet. Gynaec. Brit. Emp., 60, 834.27. ROOT, H. F., and WHITE, P. (1956), 'Diabetes Mellitus.'

Landsberger Medical Books, New York.28. HOET, J. J., GOMMERS, A., and HOET, J. P. (1958),

3rd Congress of International Diabetes Federation, Dussel-dorf, July 24.

29. HAGEN, A. (I958), Ibid.30. SHELDON, J. H. (I949), Lancet, ii, 869.31. JACKSON, W. P. U. (1954), J. din. Endocr., I4, I77.32. BOULIN, R. (1958), 3rd Congress of International Diabetes

Federation, Dusseldorf (Lancet, ii, 314).33. NEWBURGH, L. H., and CONN, J. W. (t939), J. Amer.

Ass., 112, 7.34. OGILVIE, R. F. (1935), Quart. J. Med., 4, 345.35. BEAUDOIN, R., VAN ITALLIE, T. B., and MAYER, J.

(I953), J. Clin. Nutr., I, 91.36. MAYER, J., RUSSELL, R. E., BATES, M. W., and DICKIE,

M. M. (1953), Metabolism, 2, 9.37. FOGLIA, V. G. (I944), Rev. Soc. argent. Biol., 20, 21.38. JACKSON, W. P. U. (I955), Lancet, ii, 625.39. JOSLIN, E. P., ROOT, H. F., WHITE, P., and MARBLE, A.

(1952), 'The Treatment of Diabetes Mellitus,' 9th ed.,Philadelphia)

40. WARREN, S., LE COMPTE, P. M. (I952): 'The Pathologyof Diabetes Mellitus, 3rd ed., Philadelphia.

41. AHLVIN, R. C., and BAUER, W. C. (1957), J. Dis. Child.,93, 107.

42. VAN BEEK, C. (1952), at Ist International Congress ofInternational Diabetes Federation, Leyden, Holland.

43. WOOLF, N., and JACKSON, W. P. U. (1957), 7. Path. Bact.,74, 223.

44. REIS, R. A. (x958), 3rd Congress of International DiabetesFederation, Dusseldorf, July 24.

45. YOUNG, F. G. (I95s), Brit. med.`J., ii, II67.46. ABAZA, A., VARROUD-VIAL, J., and ROMBAUTS, M.

(1953), Presse med., 6I, 495.47. HULTQUIST, G. T., and ENGFELDT, B. (1949), Acta.

Endocr. (Kbh.), 3, 365.48. JOST, A. J., and PICON, L. (1957), Symp. Soc. exp. Biol. N.Y.,

II, 228.49. LITTLE, B., VANCE, V. K., and ROSSI, E. (1958), J. clin.

Endocr., I8, 49.50. JACKSON, W. P. U., and LEWIS, B., unpublished data.5i. BEST, C. H. (I953), Ann. intern. Med., 39, 433.52. YOUNG, F. G. (I953), Recent Progr. Hormone Res., 8, 471.53. PEEL, J. H. (1958), 3rd Congress of International Diabetes

Federation, Dusseldorf, July 24.54. PENROSE, L. S. (1952), Ann. Eugen. (Camb.), i6, 378.55. MALINS, J. (1958), 3rd Congress of International Federation,

Dusseldorf, July 24.56. FUTCHER, P. H., and LONG, N. W. (1954), Bull. Johns

Hopk. Hosp., 94, 128.57. DOLGER, H. (1947), J. Amer. med. Ass., 134, 1289.58. BECKER, B. (1952), Ann. intern. Med., 37, 273.59. LAWRENCE, R. D. (I948), Brit. J. Ophthal., 325 461.60. HOET, J. P. (1954), Diabetes, 3, I.6I. MOSENTHAL, H. 0., and BOLDUAN, C. (1933), Amer.

J. med. Sci., 186, 605.62. JOSLIN, E. P., DUBLIN, L. I., and MARKS, H. H. (1936),Ibid., 191, 759; and (I937), I93, 8.63. PYKE, D. A. (1956), Lancet, i, 818.64. MURPHY, R. (I957), Conn. St. med. J., 21, 306.65. CONN, J. W. (I958), Banting Memorial Lecture, x8th Annual

Meeting American Diabetic Association, June 21.66. HOET, J. P. (I957), Symposium on ' Chronic Disease ' (WHO),Amsterdam, Sept.-Oct.67. WEISS, T. E., SEGALOFF, A., and MOORE, C. (i957),Metabolism, 6, 103.68. FAJANS, S. (1958), personal communication.69. HARRIS, S. (I924), J. Amer. med. Ass., 83, 729.70. SELTZER, H. S., FAJANS, S. S., and CONN, J. W. (i956)Diabetes, 5, 437.71. KRITZER, M. D. (x952), Med. Clin. N. Amer., 36, Is51.72. LUND, C. J., and WEESE, W. H. (1953), Amer. J. Obstet.Gynec., 65, 815.73. Brit. med. J., (1952), ii, 7I2.74. McCULLAGH, E. P., FAWELL, W. N., and LANE, F. J.(1954), 7. Amer. Ass., 156, 925.75. JACKSON, W. P. U., and WOOLF, N. (1957), Lancet, i, 614,76. BERGER, H. (I952), .. Amer. med. Ass., 148, 364.77. FAJANS, S. S., and CONN, J. W. (I954), Diabetes, 3, 296.78. WEST, K. M. (I957), Diabetes, 6, i68.79. DUNCAN, L. J. P. (I956), Quart. J. exper. Physiol., 41, 453.80. JACKSON, W. P. U., and WOOLF, N. (I958), Diabetes,7,446.81. CARDELL, B. S. (I953), J. Path. Bact., 66, 335.82. McLEAN, N., and OGILVIE, R. F. (I955), Diabetes, 4, 367

RUTHIN CASTLE, NORTH WALESA Clinic for the diagnosis and treatment of Internal Diseases (except Mental or Infectious Diseases). TheClinic is provided with a staff of doctors, nurses, technicians, modern Radiological and Physiotherapydepartments.The surroundings are beautiful. The climate is mild. There is central heating throughout. The annualrainfall is 30.5 inches, that is less than the average for England.The Fees are inclusive and vary according to the room occupied.

For particulars apply to THE SECRETARY, Ruthin Castle, North Wales.Telegrams: Castle, Ruthin Telephone: Ruthin 66

by copyright. on 17 O

ctober 2018 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.35.403.287 on 1 M

ay 1959. Dow

nloaded from