1356

SEQUENTIAL CHANGES IN SERUM INSULINCONCENTRATION DURING DEVELOPMENTOF NON-INSULIN-DEPENDENT DIABETES

MOHAMMED F. SAADDAVID J. PETTITTDAVID M. MOTT

WILLIAM C. KNOWLERROBERT G. NELSONPETER H. BENNETT

Diabetes and Arthritis Epidemiology and Clinical Diabetes andNutrition Sections, Phoenix Epidemiology and Clinical ResearchBranch, National Institute of Diabetes and Digestive and KidneyDiseases, National Institutes of Health, Phoenix, Arizona, USA

Summary Changes in serum insulin concentrationsduring deterioration of glucose tolerance

were studied in 81 Pima Indians who worsened from normalto impaired glucose tolerance (IGT); 44 who changed fromIGT to non-insulin-dependent diabetes mellitus

(NIDDM); 27 who were seen at diagnosis of NIDDM and1·4-8·5 years later; and 11 subjects who were seen at each ofthese stages. When their glucose tolerance was normal,subjects who later developed NIDDM had higher fastingand post-load insulin concentrations than controls of similarage and body mass index who did not become diabetic.Onset of IGT or NIDDM was associated with a furtherincrease in fasting insulin concentrations, although a

deterioration from IGT to NIDDM was associated withlittle change in insulin responses to oral glucose in spite ofincreased blood glucose. After the onset of NIDDM, bothfasting and post-load insulin concentrations diminished.These longitudinal data show that, as glucose toleranceworsens, insulin and glucose concentrations in individualsfollow the inverted-U-shaped relation previously reportedin cross-sectional population studies.

Introduction

HIGH, normal, and low insulin concentrations and insulinresponses to glucose have been reported in subjects withnon-insulin-dependent diabetes mellitus (NIDDM,summarised in ref 1). This variation may indicate

heterogeneity in the pathogenesis of NIDDM,23 in whichsome people have a defect that leads to diabetes in thepresence of high serum insulin concentrations, and othershave diabetes because of low serum insulin. Alternatively,high and low serum insulin concentrations might occur inthe same person at different stages of the disease.

Cross-sectional population studies of Pima Indians,4-6and other groupS/-12 have shown that fasting and glucose-stimulated insulin concentrations have an inverted-U-

shaped curve when plotted against plasma glucoseconcentrations-a pattern called "Starling’s curve of thepancreas" by DeFronzo.12 There are insufficient

prospective data from longitudinal population studies,however, to indicate whether or not an individual followssuch a pattern during development of NIDDM. We haveexamined the changes in serum insulin concentrations

during and after the onset of impaired glucose tolerance(IGT) and NIDDM among Pima Indians who werefollowed for up to 11 years.

Subjects and Methods

Data were collected during a continuing epidemiological study ofNIDDM among Pima Indians, in whom there is an extremely highprevalence of NIDDM.13 Approximately every two years, subjectsundergo a physical examination that includes a full history and have

TABLE I-DETAILS OF SUBJECTS WHO CHANGED FROM NORMAL TO

IMPAIRED GLUCOSE TOLERANCE

Median values shown (n = 44).

an oral glucose tolerance test, in which venous plasma glucose andserum insulin concentrations are measured after an overnight fastand 2 h after ingestion of a 75 g glucose-equivalent carbohydrateload.

This study includes 132 subjects, none of whom was on drugtreatment, who were followed between 1973 and 1987. Subjectswere included in one or more of the following groups: 22 men and 59women aged 8-72 years (median 23; group 1) examined whenglucose tolerance was normal and 1.6-6.7years (2-8) later when theyhad IGT; 17 men and 27 women aged 13-66 years (median 31;group 2) examined when they had IGT and at the time of diagnosisof NIDDM 1.7-10.3 years (3-1) later; 13 men and 14 women aged16-67 years (median 29; group 3) seen at the time of diagnosis ofNIDDM and 1 -4-8-5 years (2-9) later; and 2 men and 9 women aged17-52 years (median 33; group 4), examined on at least four

occasions, who initially had normal glucose tolerance, had IGT at asubsequent examination, were re-examined at or shortly after

diagnosis of NIDDM, and had a further examination after 1-4-85years (subjects in this group were followed for 5-9-10-3 years [8-3]).13 subjects were included in more than one of these groups. Forcomparison with the 11 subjects seen at all four stages, 22 controls (7men, 15 women) were selected who, at their first examination, weresimilar in age and body mass index (weight/height2), and had had atleast four examinations at similar intervals but with no evidence ofabnormal glucose tolerance. Controls were aged 16-55 years

(median 33) and were followed for 6-1-11 -2 years (8-5).Changes in body mass index, and fasting and 2 h plasma glucose

and serum insulin concentrations were studied. World Health

Organisation criteria14 for glucose tolerance were used. Plasmaglucose concentrations were measured with an automatic analyser.Serum total immunoreactive insulin concentrations were

determined by radioimmunoassay. is The interassay coefficient ofvariation of insulin was 6-8%.

Statistical analyses were done with the Statistical Package forSocial Sciences (SPSS) program.16 The Mann-Whitney test wasused for comparison between groups and the Wilcoxon paired-sample and Friedman tests for comparisons within groups.

Results

The transition from normal glucose tolerance to IGT(table I) was associated with a 71 % increase in fasting(p< 0-001) and a 182% increase in the 2-h (p<0-001)insulin concentrations. With the onset of NIDDM insubjects who initially had IGT (table II), fasting insulinconcentrations increased by 14% (p=0012), whereas 2-hinsulin concentrations fell by 33%. With an increased

1357

TABLE III-DETAILS OF SUBJECTS WITH NIDDM FOLLOWED

WITHOUT DRUG TREATMENT

Median values shown.

duration of diabetes (table in), fasting insulin decreased by33% (p=0-015) and 2-h post-load insulin concentrationsfell by 36% (p== 0-009), despite further increases in thecorresponding glucose concentrations (fig 1).Table IV shows the baseline characteristics of the subjects

who initially had normal glucose tolerance but deterioratedto IGT and then to NIDDM (group 4) compared withcontrols. When first seen, subjects who became diabetic hadsignificantly higher 2-h glucose concentrations and fastingand 2-h insulin concentrations than controls. Fig 2A showsthe body mass index, fasting and 2-h glucose concentrations,and the corresponding insulin concentrations at eachexamination during follow-up for the 11 subjects whodeveloped NIDDM. These changes are summarised in fig2B at the points when subjects had normal glucose tolerance,at the first recognition of IGT, at the first examination at or

Fig 1-Changes in median fasting (A) and 2-h (B) serum insulinconcentrations by corresponding plasma glucose concentrationsfor groups 1-4.

1=transition from normal glucose tolerance to IGT; 2 = transition fromIGT to NIDDM; 3 = progression of diabetes; 4 (broken line) = subjects seenat each stage.Enclosed area represents 95% CI of median serum insulin concentrations

grouped by plasma glucose concentrations in 2817 untreated Pima Indiansaged 15 years or older. Dotted lines represent upper limit for normal fastmgplasma glucose in A, and limits of IGT in B.

Fig 2-A. Changes in body mass index and fasting and 2-h plasmaglucose and serum insulin concentrations over time for subjectswho developed NIDDM (group 4).B. Median changes for group 4 subjects who developed NIDDM(-) and controls (- - -).

Zero on horizontal axis in A represents diagnosis of NIDDM. For subjectsin B: N = normal glucose tolerance; I = time of first diagnosis of IGT; D = atdiagnosis of diabetes; and E = at end of follow-up.

after diagnosis of NIDDM, and at the end of follow-up. Theonset of IGT was associated with an increase in fasting and2-h insulin concentrations; progression from IGT to

NIDDM was associated with a further increase in fasting,but little change in 2-h insulin concentrations; afterwards,both fasting and post-load insulin concentrations fell despitefurther increases in the corresponding glucoseconcentrations. Overall, the changes were similar to thoseobserved in the other groups during transition from onestage to the next (fig 1). For the controls, changes in fastingand 2-h glucose concentrations were not significant, butthere was an increase in insulin concentrations (p < 0-005)and body mass index (p < 0-001). However, the pattern ofchange in insulin concentrations was different from thatseen in the subjects who became diabetic (fig 2B).

Discussion

These data show a distinct pattern of change in bothfasting and post-load serum insulin concentrations inrelation to glucose concentrations during the developmentof NIDDM, and show that individuals follow an inverted-U-shaped pattern similar to that observed in cross-sectionalstudies. 4-12 Fasting and glucose-stimulated insulinconcentrations increased progressively with increased

glycaemia until glucose concentrations reached the rangethat defines IGT. Progression from IGT to NIDDM wasassociated with a further increase in fasting, but not 2-hinsulin concentrations. After the onset of NIDDM, furtherincrements in the glucose concentrations were associated

1358

with a progressive decline of fasting insulin concentrationand insulin response. High and low insulin concentrationstherefore occurred at different times in the same person.These findings indicate that insulin concentrations andresponses within individual subjects change with the degreeof glycaemia and during the course of the disease, and thathyperinsulinaemia and hypoinsulinaemia may representdifferent stages of one disease process rather than disease

heterogeneity.The pathogenesis of NIDDM is unknown and the

subject of controversy. Our findings indicate that raisedfasting and glucose-stimulated insulin concentrations

despite normal blood glucose concentrations are earlyabnormalities in subjects destined to develop NIDDM.Hyperinsulinaemia is common in populations at high risk ofNIDDM, 17-20 and Haffner et aF1 have reported thatindividuals with normal blood glucose concentrations butwho have a family history of (and presumably greatergenetic predisposition to) NIDDM have higher insulinconcentrations than those without such a family history.Furthermore, raised fasting and post-load insulinconcentrations are strong predictors of deterioration fromnormal glucose tolerance to diabetes in Pima Indians.22 Araised insulin concentration in the presence of a normalblood sugar reflects increased insulin resistance,23-25 whichappears to be a major determinant of NIDDM.26 Thetransition from normal glucose tolerance to IGT wasassociated with significant increases in fasting and 2-hinsulin concentrations: these increases probably reflectincreased insulin resistance, which appears to be mainlyresponsible for early deterioration of glucose tolerance.27The changed insulin resistance may be attributed to weightgain,28-30 aging,31,32 or both factors, although there may beother causes such as decreased physical activity.33Deterioration from IGT to NIDDM was associated with afurther increase in fasting insulin, but little change in

glucose-stimulated insulin concentrations. This findingindicates that the transition to NIDDM was caused not onlyby further increases in insulin resistance but also by inabilityof beta-cells to respond further to a glucose load. This resultis consistent with our observation that subjects with IGTwho have lower 2-h insulin concentrations are more likely toprogress to NIDDM,34 and may explain why subjects withIGT who progressed to NIDDM (group 2) had lower 2-hinsulin concentrations than those who deteriorated fromnormal glucose tolerance to IGT (group 1), because not allof the latter group progress to NIDDM. After the onset of

diabetes, both fasting and 2-h insulin concentrations

diminished, despite further increases in plasma glucoseconcentrations-an observation compatible with increasedbeta-cell failure or exhaustion. Hansen and Bodkin35 havedescribed a similar pattern of changes in insulinconcentrations in rhesus monkeys (Macaca mulatta)followed for up to 6 years until the onset of NIDDM.

Inadequate beta-cell function may be primary or

secondary. O’Rahilly et all reported abnormal pulsatileinsulin secretion in first-degree relatives of patients withNIDDM, but they had little impairment of glucosetolerance and a normal insulin response to intravenous

glucose; a primary beta-cell abnormality was suggested.Alternatively, a beta-cell defect may be secondary to chronicmild hyperglycaemia in subjects with IGT: Leahy andcolleagues3’ have shown that, in rats, mild hyperglycaemiaimpairs beta-cell secretory capacity after incompletepancreatectomy. Another possibility is that people who

become diabetic may have a small beta-cell mass that makes

them less able to compensate for increased insulinresistance.38 Cahill39 suggested that subjects with

genetically predetermined NIDDM might have a morerapid age-related beta-cell death.From these results and other data ’15-27, 34,40 we suggest the

following sequence of events during the onset of NIDDMin Pima Indians. The primary abnormality appears tobe increased insulin resistance, probably geneticallydetermined .25 40 Beta-cells react to the increased insulinresistance by increased insulin secretion, in both the fastingstate and in response to a glucose load, to maintain normalblood glucose concentrations. However, with aging andincreased obesity-and possibly other factors-insulinresistance increases further, and IGT occurs. Progression toNIDDM is associated with a further increase in insulinresistance, a failure of beta-cells to increase their response toa glucose load, or both abnormalities. The relative severityof insulin resistance and impaired beta-cell functiondetermines if and when NIDDM occurs, and the severity ofhyperglycaemia. After the onset of NIDDM there is afurther reduction in beta-cell function and eventually astriking fall in both basal and glucose-induced insulinsecretion: the reduction in blood insulin concentrations maybe underestimated because conventional methods of insulin

radioimmunoassay also measure proinsulin, which may beraised disproportionately in many patients with NIDDM.41Treatment of hyperglycaemia partially restores the defectivebeta-cell response,42-44 but this defect eventually becomesirreversible45 and basal insulin secretion falls.

We thank the Gila River Indian Community for its cooperation; themedical, laboratory, and data processing staff of Phoenix Epidemiology andClinical Research Branch for technical assistance; and Ms Charlene K. Gishiefor secretarial assistance.

Correspondence should be addressed to M. F. S., Diabetes and ArthritisEpidemiology Section, NIDDK, 1550 East Indian School Road, Phoenix,Arizona 85014, USA.

REFERENCES

1. DeFronzo RA, Ferrannini E. The pathogenesis of non-insulin-dependent diabetes.An update. Medicine 1982; 61: 125-40.

2. Fajans SS, Clouner MC, Crowther RL. Clinical and etiologic heterogeneity ofidiopathic diabetes mellitus. Diabetes 1978; 27: 1112-25.

3 Fajans SS. Heterogeneity of insulin secretion m type II diabetes. Diabetes Metab Rev1986, 2: 347-61.

4. Savage PJ, Dippe SE, Bennett PH, et al. Hyperinsulinemia and hypoinsulinemia.Insulin responses to oral carbohydrate over a wide spectrum of glucose toleranceDiabetes 1975; 24: 362-68.

5. Bennett PH, Knowler WC, Pettitt DJ, Carraher MJ, Vasquez B. Longitudinal studiesof the development of diabetes in the Pima Indians. In: Eschwege E, ed. Advancesin diabetes epidemiology, INSERM Symposium 22. Amsterdam: Elsevier, 1982.65-74.

6. Bogardus C, Lillioja S, Howard BV, Reaven G, Mott D. Relationships betweeninsulin secretion, insulin action, and fasting plasma glucose concentration innondiabetic and noninsulin dependent diabetic subjects. J Clin Invest 1984, 74:1238-46

7. Reaven G, Miller R. Study of the relationship between glucose and insulin responsesto an oral glucose load in man. Diabetes 1968; 17: 560-69.

8. Welborn TA, Stenhouse NS, Johnstone CG. Factors determining serum insulinresponse m a population sample. Diabetologia 1969; 5: 263-66.

9. Zimmet P, Whitehouse S, Alford F, Chisholm D. The relationship of insulin responseto a glucose stimulus over a wide range of glucose tolerance. Diabetologia 1978; 15:23-27.

10. Martin FIR, Wyatt GB, Griew AR, Haurahelia M, Higginbotham L. Diabetesmellitus in urban and rural communities in Papua, New Guinea: studies ofprevalence and plasma insulin. Diabetologia 1980; 18: 369-74

11. Zimmet P, Whitehouse S, Kiss J. Ethnic variability in the plasma insulin response tooral glucose in Polynesian and Micronesian subjects. Diabetes 1979; 28: 624-28

12. DeFronzo RA The triumvirate. B-cell, muscle, liver A collusioon responsible forNIDDM. Diabetes 1988; 37: 667-87.

13. Knowler WC, Bennett PH, Hamman RF, Miller M. Diabetes incidence andprevalence in Pima Indians a 19-fold greater incidence than in RochesterMinnesota Am J Epidemiol 1979, 108: 497-505.

14. Report of a WHO Study Group. Diabetes mellitus. WHO Tech Rep Ser 1985, 727:9-17.

1359

15 Herbert V, Lau K, Gottlieb CW, Bleicher SJ. Coated charcoal immunoassay ofinsulin. J Clin Endocrinol Metab 1965; 25: 1375-84.

16. SPSSX. User’s Guide, 2nd ed. Chicago: SPSS, 1986: 800-35.17. Aronoff SL, Bennett PH, Gorden P, Rushforth N, Miller M. Unexplained

hyperinsulinemia in normal and "prediabetic" Pima Indians compared withnormal Caucasians. Diabetes 1977; 26: 827-40

18. Balkau B, King H, Zimmet P, Raper LR. Factors associated with the development ofdiabetes in the Micronesian population of Nauru. Am J Epidemiol 1985; 122:594-605.

19. Haffner SM, Stem MP, Hazuda HY, Pugh JA, Patterson JK. Hyperinsulinemia in apopulation at a high risk for non-insulin-dependent diabetes mellitus. N Engl JMed 1986, 315: 220-24.

20. O’Dea K, Traianedes K, Hopper JL, Larkins RG. Impaired glucose tolerance,hyperinsulinemia, and hypertriglyceridemia in Australian Aborigines from thedesert. Diabetes Care 1988, 11: 23-29.

21. Haffner SM, Stem MP, Hazuda HP, Mitchell BD, Patterson JK. Increased insulinconcentrations in nondiabetic offspring of diabetic parents. N Engl J Med 1988;319: 1297-301.

22. Knowler WC, Bennett PH. Serum insulin concentrations predict change in oralglucose tolerance. Diabetes 1983; 32: 46A.

23 Nagulesparan M, Savage PJ, Unger RH, Bennett PH. A simplified method usingsomatostatin to assess in vivo insulin resistance over a range of obesity. Diabetes1979; 28: 980-83.

24. Hollenbeck CB, Chen N, Chen Y-DI, Reaven GM. Relationship between the plasmainsulin response to oral glucose and insulin-stimulated glucose utilization in normalsubjects. Diabetes 1984; 33: 460-63.

25. Lillioja S, Mott DM, Zawadzki JK, et al. In vivo insulin action is familial characteristicm non-diabetic Pima Indians. Diabetes 1987; 36: 1329-35.

26. Bogardus C, Lillioja S, Foley J, et al. Insulin resistance predicts the development ofnon-insulin-dependent diabetes mellitus in Pima Indians. Diabetes 1987; 36: 47A.

27 Lillioja S, Mott DM, Howard BV, et al. Impaired glucose tolerance as a disorder ofinsulin action: longitudinal and cross-sectional studies in Pima Indians. N Engl JMed 1988; 318: 1217-25.

28. Kolterman OG, Insel J, Saekow M, Olefsky JM. Mechanisms of insulin resistance inhuman obesity: evidence for receptor and post-receptor defects. J Clin Invest 1980,65: 1272-84.

29 Kashiwagi A, Bogardus C, Lillioja S, et al. In vitro insensitivity of glucose transportand antilipolysis to insulin due to receptor and post-receptor abnormalities in obesePima Indians with normal glucose tolerance. Metabolism 1984; 33: 772-77.

30. Lillioja S, Young AA, Cutler CL, et al. Skeletal muscle capillary density and fiber typeare possible determinants of in vivo insulin resistance in man. J Clin Invest 1987; 80:415-24.

31. Fink RI, Kolterman OG, Griffin J, Olefsky JF. Mechanisms of insulin resistance ofaging. J Clin Invest 1983; 71: 1523-35.

32. Rowe JW, Minaker KL, Pallota JA, Flier JS. Characterization of the insulin resistanceof aging. J Clin Invest 1983; 71: 1581-87.

33. Yki-Jarvmen H, Kolvisto VA. Effects of body composition on insulin sensitivity.Diabetes 1983; 32: 965-69.

34. Saad MF, Knowler WC, Pettitt DJ, Nelson RG, Mott DM, Bennett PH. The naturalhistory of impaired glucose tolerance in the Pima Indians. N Engl J Med 1988; 319:1500-06.

35. Hansen BC, Bodkin NL. Heterogeneity of insulin responses: phases leading to type 2(non-insulin-dependent) diabetes mellitus in the rhesus monkey. Diabetologia1986; 29: 713-19

36. O’Rahilly S, Turner RC, Matthews DR. Impaired pulsatile secretion of insulin inrelatives of patients with non-insulin-dependent diabetes. N Engl J Med 1988; 318:1225-30.

37. Leahy JL, Bonner-Weir S, Weir GC. Minimal chronic hyperglycemia is a criticaldeterminant of impaired insulin secretion after an incomplete pancreatectomy.J Clin Invest 1988; 41: 1407-14.

38. Weir GC, Leahy JL, Bonner-Weir S. Experimental reduction of B-cell massimplications for the pathogenesis of diabetes. Diabetes Metabol Rev 1986; 2: 125-61.

39 Cahill GF Jr. Beta-cell deficiency, insulin resistance or both? N Engl J Med 1988; 318:1268-70.

40. Bogardus C, Lillioja S, Nyomba BL, et al Evidence for a single gene, co-dominantmode of inheritance of insulin resistance in Pima Indians. Diabetes 1988; 37: 91A.

41. Temple RC, Carrington CA, Luzio SD, et al. Insulin deficiency in non-insulin-dependent diabetes. Lancet 1989; r 293-95.

42. Savage PJ, Bennion LJ, Flock EV, et al Diet-induced improvement of abnormalitiesin insulin and glucagon secretion and in insulin receptor binding m diabetesmellitus. J Clin Endocrinol Metab 1979; 48: 999-1007

43. Kosaka K, Kuzuya T, Akanuma Y, Hagura R. Increase in insulin response aftertreatment of overt maturity-onset diabetes is independent of the mode of treatmentDiabetologia 1980; 18: 23-28.

44. Stanik S, Marcus R. Insulin secretion improves following control of plasma glucose inseverely hyperglycemic obese patients. Metabolism 1980; 29: 346-50.

45. Nagulesparan M, Savage PJ, Bennion LJ, Unger RH, Bennett PH. Diminished effectof caloric restriction on control of hyperglycemia with increasing known duration oftype II diabetes mellitus. J Clin Endocricol Metab 1981; 53: 560-68.

Reviews of Books

Mobilizing against AIDS

2MM aJ eM/aJ !fMM.&mdash;Eve K. Nichols. CambridgeRevised and enlarged edition.-Eve K. Nichols. Cambridge(Mass) and London: Harvard University Press. 1989. Pp 387.$12.95. ISBN 0-674577620.

THIS impressive book is the official effort of the Instituteof Medicine in the United States to bring an accurateunderstanding about AIDS to the public at large. In ninedense but interesting chapters, Eve Nichols describes theworldwide scope of the epidemic, shows how the diseasedevelops, and reviews the history of the search for the cause.The chapters unravelling the virological and immunologicalfacts of AIDS are models of clarity, presenting complexideas and data with precision. By using carefully definedscientific terminology with simplicity and directness, sheleads even a biologically untutored reader towards

understanding.The section dealing with prevention by education and

other public health methods of control leaves one depressedby the magnitude of the problems. Current drug therapiesare described and the obstacles to drug and vaccinedevelopment are explained in detail; this is not an optimisticchapter. The final chapters on the impact of AIDS onpatients and their families and friends, on society, and on themedical care system are sobering in the extreme.Valuable appendices bring together data such as the 1987

CDC case definition, information on available tests, andpublic health guidelines on the spread of AIDS in medicalsettings and the workplace.

For general readers without medical or biologicalbackground, this book should serve to put AIDS into anaccurate perspective; it may be particularly useful to

decision-makers in government and industry who need tounderstand how the illness will affect their clients,employees, or customers. Despite the author’s skill in

translating biological ideas into ordinary language, graspingthis material will take a considerable effort. It cannot berecommended for light vacation reading. For physicians,especially those who have not yet been directly involvedwith AIDS, the book will provide a valuable generaloverview of the illness and its medical and societal

implications.Berkeley Pediatric Medical Group,1650 Walnut Street,Berkeley, CA 94709, USA ELMER R. GROSSMAN

Arthritis and Allied Conditions

11 th edition.-Edited by Daniel J. McCarty. Beckenham: Quest(Lea and Febiger). 1989. Pp 2045.$150. ISBN 0-812111230.

BASIC research into inflammatory joint diseases is

proceeding at a prodigious rate, and all rheumatologistsneed to comprehend the important developments.Unfortunately, the scientific section of the latest edition ofArthritis and Allied Conditions is not a success. Byapproaching the problem in terms of specific cells

(neutrophils, lymphocytes, mast cells, and so on) andmolecules (complement, immunoglobulins) it fails to reflectthe dynamic inter-reactions, subtlety, and beauty of theinflammatory cascade. Many of the contributors in this