B E H A V I O R A L D I A B E T E S S E R I E S A R T I C L E

Stress and Diabetes MellitusRICHARD S. SURWIT, PHDMARK S. SCHNEIDER, PHDMARK N. FEINGLOS, MD

Stress is a potential contributor to chronic hyperglycemia in diabetes. Stress has longbeen shown to have major effects on metabolic activity. Energy mobilization is aprimary result of the fight or flight response. Stress stimulates the release of varioushormones, which can result in elevated blood glucose levels. Although this is ofadaptive importance in a healthy organism, in diabetes, as a result of the relative orabsolute lack of insulin, stress-induced increases in glucose cannot be metabolizedproperly. Furthermore, regulation of these stress hormones may be abnormal indiabetes. However, evidence characterizing the effects of stress in type I diabetes iscontradictory. Although some retrospective human studies have suggested that stresscan precipitate type I diabetes, animal studies have shown that stressors of variouskinds can precipitate—or prevent—various experimental models of the disease.Human studies have shown that stress can stimulate hyperglycemia, hypoglycemia,or have no affect at all on glycemic status in established diabetes. Much of thisconfusion may be attributable to the presence of autonomic neuropathy, common intype I diabetes. In contrast, more consistent evidence supports the role of stress intype II diabetes. Although human studies on the role of stress in the onset and courseof type II diabetes are few, a large body of animal study supports the notion that stressreliably produces hyperglycemia in this form of the disease. Furthermore, there ismounting evidence of autonomic contributions to the pathophysiology of this con-dition in both animals and humans.

£ £ j j ^ ut if the degenerate, or de-B ^ praved nervous liquor dothwJ continually flow into the

blood, it produces sometimes the un-bloody dysentery, such as we have alreadydescribed, sometimes the diabetes...."

Thomas WillisPharmaceutis Rationalis, 1679

The notion that stress could con-tribute to the etiology of diabetes melli-tus has a long history in medicine—asthe 17th century physician Willis noted(1). By the 19th century, the role of stressin the etiology of diabetes had becomeestablished firmly in the medical litera-ture. Henry Maudsley, considered by

FROM THE DEPARTMENTS OF PSYCHIATRY AND MEDICINE, DUKE UNIVERSITY MEDICAL CENTER, DURHAM,

NORTH CAROLINA.

ADDRESS CORRESPONDENCE AND REPRINT REQUESTS TO RICHARD S. SURWIT, PHD, BOX 3842, DUKE

UNIVERSITY MEDICAL CENTER, DURHAM, NC 27710.

RECEIVED FOR PUBLICATION 26 AUGUST 1991 AND ACCEPTED IN REVISED FORM 31 MARCH 1992.

TYPE 1 DIABETES, INSULIN-DEPENDENT DIABETES MELLITUS; TYPE II DIABETES, NON-INSULIN-DEPENDENT

DIABETES MELLITUS; G H , GROWTH HORMONE; F F A , FREE FATTY ACID; E P I , EPINEPHRINE; S T Z , STREPTO-

ZOCIN; NE, NOREPINEPHRINE.

THIS ARTICLE IS ONE OF A SERIES PRESENTED AT THE MEETING ON THE BEHAVIORAL ASPECTS OF DIABETES

MELLITUS.

many to be the founder of modern psy-chiatry, wrote, "This we know: that dia-betes is sometimes caused in man bymental anxiety . . . (2)." As Willis hadbefore him, Maudsley observed that dia-betes often followed the occurrence of asudden trauma. He reported the story ofa military officer, who, upon discoveringthat his wife was having an affair, imme-diately developed the disease. No less afigure than the great William Osier alsosubscribed to the notion that stress wasinvolved in the etiology of some types ofdiabetes. In his landmark Textbook ofMedicine, Osier differentiated betweentrue diabetes and the less severe diabetesof obesity, probably analogous to what iscalled type II diabetes today: "In truediabetes, instances of cure are rare. Onthe other hand, the transient or intermit-tent glycosuria met with in stout over-feeders, or in persons who have under-gone a severe mental strain, is veryamenable to treatment (3)."

By the beginning of the 20th cen-tury, laboratory studies began to replaceastute clinical observation as the princi-pal source of medical knowledge. Com-menting on existing clinical literature,Walter B. Cannon made the first appealfor the experimental study of how stressaffects diabetes. "Although clinical evi-dence thus indicates an emotional originof some cases of diabetes and glycos-uria," Cannon wrote, "the intricacies ofexistence and the complications of dis-ease in human beings throw some doubton the value of that evidence. . . .it isdesirable, therefore, that the question ofan emotional glycosuria be tested undersimpler and more controllable condi-tions." To study the problem experimen-tally, Cannon (4) provoked stress-in-duced hyperglycemia in normal cats. Inthis early investigation, 12 cats were con-fined in a holder for varying lengths oftime, dependent on each animal's reac-tion to this novel situation. The cats weregiven a large quantity of water by stom-ach tube, and urine was drainedpromptly. In all cases, sugar was absent

DIABETES CARE, VOLUME 15, NUMBER 10, OCTOBER 1992 1413

Stress and diabetes

from the urine before the animal becameexcited, but the stress intervention in-variably resulted in glycosuria. Cannonobserved an apparent relationship be-tween the animal's emotional state andthe onset of hyperglycemia. Specifically,animals that appeared frightened or en-raged developed glycosuria more quicklythan animals that responded to the con-finement in a calm manner.

Cannon attributed this effect tothe fight or flight response, which in-cludes sympathetic discharge and eleva-tions in circulating levels of catechol-amines, glucocorticoids, and GH. Thenet effect of this response is energy mo-bilization. In that glucoregulation iscompromised in diabetic individuals, theenergy-mobilizing effects of stress can bedeleterious to blood glucose control in adiabetic patient (5,6). Therefore, the ex-tent to which environmental stress andother behavioral variables contribute toblood glucose control is theoretically im-portant in the clinical management ofthis disorder.

It is now widely accepted that theautonomic/adrenocortical systems play amajor role in the regulation of carbohy-drate metabolism (Fig. 1). The effects ofthe autonomic nervous system on insulinaction is both facilatory and inhibitory.Branches of the parasympathetic rightvagus nerve innervate the pancreatic is-lets, and stimulation of the right vaguscauses increased insulin secretion. Ad-renergic stimulation of pancreatic isletcells can lead to either facilitation or in-hibition of insulin secretion, fi-adrener-gic stimulation at low levels is facilatoryto insulin output, whereas high levels of3-adrenergic stimulation or a-2 adren-ergic stimulation is inhibitory, fi-adren-ergic stimulation also stimulates gluca-gon release from the pancreatic a-cells.Glucagon, in turn, stimulates glucoseproduction in the liver. Thus, sympa-thetic and parasympathetic innervationof the pancreas modulates the normalregulation of carbohydrate metabolism(7-9). p-adrenergic stimulation alsopromotes the conversion of glycogen to

Figure 1—Schematic of pathways by whichthe central nervous system (CNS) can influenceglucose metabolism. Adrenergic (a and (3), cho-linergic (muscarinic), and opioidergic pathwaysare shown. Adrenal cortical pathways are notshown.

glucose in the liver and fat to FFAs inadipose tissue. FFAs further are metabo-lized to ketoacids in the liver. CirculatingEPI can stimulate glycogenolysis (7),whereas vagal stimulation can inhibit he-patic glucose production (10). Thus, theautonomic nervous system has both di-rect neural and indirect hormonal con-trol pathways in the regulation of glucosemetabolism (8).

Stressful stimuli can activate sev-eral other neuroendocrine responses thatcan result in elevated blood glucose lev-els. Activation of the hypothalamic- pitu-itary- adrenocortical axis causes release ofincreased amounts of glucocorticoids,enhances gluconeogenesis in the liver,and diminishes cellular glucose uptake.Stress-induced release of GH also candecrease glucose uptake, and fi-endor-phin can suppress insulin secretion andelevate glucose levels (11). Therefore,stressful stimuli can impact on glucoselevels through numerous pathways. Theadaptive benefit of stress-induced energymobilization in healthy, nondiabetic in-dividuals is obvious. However, in diabe-tes mellitus, where glucose metabolism iscompromised, these stress effects can beproblematic. Finally, elevated blood glu-cose levels can, by themselves, eventuallyimpair the pancreas' ability to respond toa glucose stimulus (12). Thus, glucose

toxicity resulting from chronic, intermit-tent, stress-induced elevations in bloodglucose could produce a permanent ef-fect on pancreatic secretory ability.

THE ROLE OF STRESS IN THEONSET OF TYPE IDIABETES— Animal research hasprovided some objective evidence to sug-gest that stress affects the onset of type Idiabetes. Animals that were partiallypancreatectomized surgically have beenshown to develop either transient or per-manent diabetes after restraint stress(13). Although these animals do not de-velop diabetes spontaneously, restraintstress produced transient or permanentdiabetes in 25-28% of pancreatecto-mized animals, but none of the controlanimals developed diabetes. Some ani-mals who were not altered surgically be-came hyperglycemic after the stressor,but none became diabetic. More re-cently, chemical pancreatectomy withf$-cell cytotoxins, such as alloxan andSTZ, have been used instead of surgicalprocedures. By controlling the dosage ofthese cytotoxic agents, partial or com-plete destruction of pancreatic f$-cellmass can be produced, mimicking theclinical picture of type II and type I dia-betes, respectively. Huang et al. (14)found that light-shock stimulation couldinhibit the development of STZ-induceddiabetes in young mice that received asingle dose of STZ. Although the mech-anism of this effect was not defined,other researchers have found that admin-istration of exogenous steroids can in-hibit the development of another STZ-induced model of diabetes (15). It ispossible, therefore, that the protective ef-fect of shock on the development of di-abetes observed in animals treated with asingle dose of STZ may have been medi-ated through the adrenal corticotrophiceffects of stress.

The diabetes-prone BB Wistar rat(16) provides a genetic model for Type Idiabetes. A significant percentage ofthese animals spontaneously develop anautoimmune insulitis that results in dia-

1414 DIABETES CASE, VOLUME 15, NUMBER 10, OCTOBER 1992

Surwit, Schneider, and Feinglos

betes by the time they are 5 mo old. Theeffects of stress on BB rats have beendemonstrated by Carter et al. (17). Acombination of behavioral stressors,such as restraint and crowding, werefound to lower the age of onset of diabe-tes. Lehman et al. (18) found that agreater percentage of animals became di-abetic after stress, compared with un-stressed controls. Because BB rats possessother endocrine and immune abnormal-ities, one must be cautious in generaliz-ing these findings to humans.

Stress also has been suspected toplay a role in the onset of type I diabetesin humans. Studies have shown that di-abetic patients are more likely to suffer amajor family loss before the onset ofsymptoms (19-21). However, thesestudies tend to be poorly controlled,and/or they rely on recall of specific lifeevents. Robinson and Fuller (19) didcompare diabetic patients with nondia-betic siblings of similar age and amatched neighborhood control group.The diabetic subjects had significantlymore severe life events within the 3 yrbefore diagnosis than either controlgroup. Severe life events were definedbased on the degree of short-term orlong-term threat to the subject. Thisstudy is limited by its small sample size,and it fails to give specific examples ofsevere life events. Although these studiesare far from conclusive, it must be notedthat only 50% of identical twins are con-cordant for type I diabetes. Althoughboth show evidence of autoimmune ab-normalities, the disease develops in only50% of the pairs. It has been postulated,therefore, that some environmental stim-ulus is necessary for the overt expressionof the disease. Therefore, stress may af-fect the onset of type I diabetes by di-rectly or indirectly triggering this au-toimmune abnormality. However,conclusive evidence for this mechanismis lacking.

STRESS AND GLUCOSE CONTROLIN TYPE I DIABETES— The litera-ture on the relationship between stress,

autonomic arousal, and glucose controlin diabetes is more extensive. Lee et al.(22) have demonstrated that not all ratswith chemically induced diabetes re-spond to stress in the same way. Aftereither alloxan or STZ injections, ~50%of the animals demonstrated elevatedbaseline plasma EPI levels (reactive re-sponders) whereas the other 50% (non-reactive responders) did not differ fromcontrols in terms of plasma EPI. Afterfootshock, the reactive animals hadplasma EPI levels >5460 pM, whereasthe nonreactive and control rats were<3276 pM. The three groups did notdiffer in terms of NE response to stress.Reactive and nonreactive animals had el-evated blood glucose at baseline (>11.2raM) relative to controls (<5.6 mM), andall groups showed increased blood glu-cose in response to stress. This suggeststhat the reactive animals have an in-creased secretion of EPI from the adrenalmedulla rather than a decreased rate ofremoval of EPI from the blood (22). Onthe other hand, Bellush and Rowland

(23) found that all STZ rats had reliablyhigher NE than controls, both before andafter stress. EPI was found to increase indiabetic animals after stress and decreasein controls. In this study, 24-h urinarycatecholamine excretion was used ratherthan plasma levels. The authors concludethat the elevated NE levels in the diabeticrats suggest greater sympathetic activityin those animals. These studies highlightthe complex nature of the relationshipbetween environmental stimulation anddiabetes in animals rendered susceptibleto the disease.

Models that produce artificiallyinduced diabetes do have some prob-lems. First, pancreatectomized animalsundergo potentially stressful proceduresthat differ from type I diabetes onset inhumans (24). The stress of the suddenonset of the condition in these animalmodels may lead to changes in catechol-amines, independent of the effects of ex-perimental stressors on the condition it-self (22). Second, the animals used inthese studies often have chronic hyper-

glycemia that is not being controlled byinsulin. Therefore, it may not be appro-priate to generalize these findings to hu-mans with type I diabetes, who often arein reasonable glycemic control with in-sulin.

In humans, several physical stres-sors such as illness, trauma, or rapidmetabolic changes have been shown tocause hyperglycemia and eventual ke-toacidosis in type I diabetes (25).McLesky et al. (26) found a clear hyper-glycemic response in both type I andtype II diabetic patients during a surgicalstress. Furthermore, insulin-dependentdiabetic children have been shown todemonstrate elevated blood glucose andmore rapid ketone release after EPI infu-sion compared with normal children(27).

The experimental study of the ef-fects of psychological stress on glucosemetabolism in diabetic patients was un-dertaken first by Hinkle et al. (28-30).Their studies of diabetic patients demon-strated increases in blood glucose andketones after stressful psychiatric inter-views. However, their work was poorlycontrolled and difficult to interpret. Van-denbergh et al. (31) examined the impactof hypnotically induced emotion on pre-dominantly IDDM subjects, in a con-trolled fashion. Coincident with nonsig-nificant increases in plasma FFAs, theyobserved decreases in blood glucose.These findings were replicated in a sub-sequent study (32). However, these re-searchers did not fully documentwhether or not their subjects had endog-enous insulin reserves. Recent negativelife events also have been shown to havean influence on diabetic control. Chaseand Jackson (33) found that life stress, asdefined by a list of events requiring psy-chosocial readjustment, is correlatedwith both GHb and blood glucose. Thissuggests that stress has an impact onboth long-term and short-term indexesof glycemic control. Brand et al. (34), onthe other hand, found life stress to becorrelated only with urinary ketone lev-els in males. As with most studies in this

DIABETES CARE, VOLUME 15, NUMBER 10, OCTOBER 1992 1415

Stress and diabetes

area, specific examples of stressors werenot provided. These studies used differ-ent measures of life stress and neitherinvestigated whether the findings suggesta direct effect of stress on glycemic con-trol or an indirect effect based on poorercompliance.

Not all studies have demon-strated that laboratory stress leads toblood glucose changes in type I diabeticpatients. In one study, type I diabeticpatients in good glycemic control werecompared with individuals in an acutestate of insulin insufficiency. After men-tal-arithmetic and public-speaking stres-sors, neither group showed significantchanges in blood glucose levels (35). Be-cause artificially induced insulin defi-ciency may not be comparable withchronic poor glycemic control, Gilbert etal. (36) compared patients in good gly-cemic control with those who had a his-tory of poor control as measured byGHb. Neither group showed a significantchange in blood glucose after a publicspeaking stressor, although the poorlycontrolled diabetic subjects had higherpoststress urine glucose levels.

Other studies have found idio-syncratic changes in blood glucose afterlaboratory stressors. Bradley (37) re-ported that noise stress increased bloodglucose in initially hyperglycemic dia-betic subjects and decreased blood glu-cose in initially hypoglycemic diabeticsubjects. Some have found that mentalarithmetic can produce either increasesor decreases in blood glucose (38) intype 1 diabetic patients, and that the di-rection of blood glucose change is idio-syncratic. Stabler et al. (39) examinedthe relationship between glycemic re-sponse to stress and behavior patterns oftype A personality individuals. Theyfound that type A children showed ele-vated blood glucose levels after playing astressful video game, whereas type Bchildren exhibited decreased blood glu-cose levels. Although the differences inblood glucose were small and probablynot clinically significant, type A childrenhad higher GHb levels than type B chil-

dren. A subsequent study failed to repli-cate the GHb differences between type Aand type B children, although glucosereactivity to stress was not assessed (40).Thus, some of the studies investigatingthe impact of psychological stress onblood glucose in human diabetes reportthat stress has a hyperglycemic effect,whereas others find that the response isidiosyncratic, with some patients show-ing hyperglycemia and some showinghypoglycemia in response to stressfulstimuli. Stabler et al.'s study (39) is no-table in that it suggests that these differ-ences in response to stress could be re-lated to some personality variable. Theissue of how individuals differ in meta-bolic response to stress has significantclinical implications and should be stud-ied further.

Although the effects of stressfulbehavioral manipulations on blood glu-cose are somewhat contradictory, thedata on the role of stress hormones (cat-echolamines, cortisol, GH) in the devel-opment of hyperglycemia and ketosis intype I diabetes appears to be more con-sistent. Fernqvist et al. (41) studied theeffects of different doses of EP1 on theabsorption of subcutaneous insulin. Bothhealthy, nondiabetic and diabetic sub-jects exhibited decreased absorption ofinsulin, even though subcutaneousblood flow remained stable or increased.At high levels of EPI (0.3 nmol • kg"1 •min"1 intended to mimic moderatephysical stress, insulin absorption wasretarded by as much as 50%. One prob-lem with this study was that not all of thesubjects may have had Type I diabetes,so some may have had an endogenousinsulin supply. In a series of well-controlled studies, Sherwin et al. (42)examined the effects on plasma glucoseof infusions of EPI, glucagon, and corti-sol, both alone and in combination innormal subjects and type I diabetic pa-tients. The type I diabetic patients re-ceived insulin infusions for several hoursbefore injection of the hormones. AfterEPI infusion, the elevation of blood glu-cose was 4 -6 times greater in type I

diabetic patients compared with normalcontrol subjects. This increase was sus-tained in type I diabetic patients over a5-h period, but lasted <2 h in the con-trol subjects. Cortisol infusions pro-duced a 5-7 times greater increase inblood glucose in the type I diabetic pa-tients compared with normal controlsubjects. An increase in hepatic glucoseproduction was observed only in thetype I diabetic patients.

The apparently contradictory re-sults in studies on the effects of stress intype I diabetic patients and the moreconsistent results on the stress hormone-induced hyperglycemia probably are at-tributable to several factors. First, theterm stress is used to describe an enor-mous variety of both experimental stim-uli (e.g., noise) and responses (e.g., men-tal arithmetic). In the studies just cited,no two groups of researchers used thesame stress, hence, the disparity of re-sults is not surprising. Furthermore,many of these studies did not describethe nature of their subjects' diabetescarefully. The effects of a given environ-mental stimulus on a patient with someendogenous insulin could be quite dif-ferent than if applied to a patient withoutendogenous insulin. Finally, disruptionsin autonomic nervous system activity,because of diabetic neuropathy, can leadto decreased sympathetic responses indiabetic patients (43). In patients withknown autonomic neuropathy, mental-arithmetic stress does not produce thetypical changes in skin temperature andheart rate that are found in both diabeticpatients without neuropathy and nondi-abetic control subjects (44). Becausethese sympathetic nervous system de-fects are common in diabetes, Cryer (43)has argued that increased sympatheticnervous system activity is not likely to beresponsible for hyperglycemia in type Idiabetes. Variations in autonomic statusamong diabetic patients may account forthe different stress effects found in vari-ous studies and also explain why exoge-nously administered stress hormones

1416 DIABETES CARE, VOLUME 15, NUMBER 10, OCTOBER 1992

Surwit, Schneider, and Feinglos

produce more reliable hyperglycemic ef-fects.

THE ROLE OF STRESS IN THEONSET OF TYPE IIDIABETES— Although it is nowknown that type 1 diabetes results fromautoimmune destruction of the pancre-atic p-cells, the pathophysiology of typeII diabetes remains obscure. Conven-tional theories suggest that type II diabe-tes is caused by either a primary defect inthe p-cell, making it less responsive toglucose stimulation, or by the severe in-sulin resistance that eventually exhaustsP-cell function (45). However, attemptsto find such a defect in the P-cell or sucha mechanism for insulin resistance in so-matic cells has been frustrating. Exami-nation of genes that code for insulin pro-duction or insulin-receptor expressionhave failed to identify defects in thesefunctions (46).

Over the past 10 years, there hasbeen speculation that the autonomic ner-vous system is involved in the patho-physiology of type II diabetes (11,47-50). This possibility was foreseen byClaude Bernard, who found that hyper-glycemia could be produced in normal,nondiabetic rabbits by lesioning the areaof the hypothalamus. More recently, sev-eral researchers have shown that hyper-glycemia can be produced by chemicalstimulation of the brain with morphineand by a variety of endogenous neu-ropeptides, and that it can be abolishedby bilateral adrenalectomy (11,51). Hy-perglycemia also has been found to occurfrom a slow intravenous infusion of EPI(52) and from the type of stress thatresults in prolonged sympathetic dis-charge (53). Autonomic activity thatleads to metabolic decompensationcould be stimulated by stress (50) or bythe effects of dietary fat and carbohydrateon sympathetic outflow (54-55).

One of the first observations thatstress could contribute to the expressionof hyperglycemia in an animal model ofspontaneously occurring type II diabeteswas made during metabolic studies of

the sand rat (psammomys obesus) (57).The sand rat is a north African rodentthat eats an exclusively low calorie diet ofsucculent plants in its natural habitat.Sand rats that are maintained on a low-calorie, low-carbohydrate diet do not de-velop diabetes. However, when they arefed laboratory chow and allowed to be-come obese, a significant percentage ofthe animals develop an analog of type IIdiabetes (56). Mikat et al. (57) haveshown that stress, diet, and obesity eachmay play a role in the expression of hy-perglycemia in these animals. The re-searchers maintained sand rats on a low-calorie, low-carbohydrate diet ofvegetables and saline, so that they re-mained euglycemic. Glucose or salinethen was administered to rats eitherthrough an esophageal tube or intraper-itoneally by injection. Similar procedureswere carried out on a group of Sprague-Dawley rats. Blood samples were drawnfrom all of the animals at 30- and 120-min intervals, and analyzed for glucoseand insulin. Sand rats that received glu-cose via an intraperitoneal injection (notintubated) showed normal glucose toler-ance values. However, sand rats withesophageal intubation showed the ab-normal glucose tolerance that would beconsidered typical of diabetes. In con-trast, in the Sprague-Dawley rats, intuba-tion did not alter normal glucose toler-ance values. Thus, it appears that stress,such as that of intubation, precipitatesglucose intolerance even in lean, eugly-cemic sand rats genetically predisposedtowards developing diabetes.

The genetically obese mouse(C57BL76J, ob/ob) is a more commonlyused model of type II diabetes. This an-imal is a good model of type II diabetesbecause it is characterized by a syndromeof obesity, hyperinsulinemia, insulin re-sistance, hyperglycemia, and glucose in-tolerance (56). However, some contro-versy has existed over the degree towhich the obese mouse is hyperglyce-mic. Different laboratories have reportedvarying basal plasma glucose levels,ranging from 7.28 mM to >16.8 mM

CONTROL STRESS

~ 700-

30-

Figure 2—Mean ± SE effects of stress onplasma glucose and insulin in C57BL/6J ob/obmice and their lean littermates. From Surwit etal. (61). © by the American Diabetes Associa-tion.

(58-60). Surwit et al. (61) have shownthat the degree of hyperglycemia is de-pendent, in part, on whether the animalis exposed to stressful environmentalstimuli. In one experiment, blood sam-ples were drawn from obese and leanmice after either a rest period or expo-sure to stress. Stress consisted of restraintin a wire-mesh cage for 60 min, punctu-ated by a 5-min period of shaking. Bothlean and obese animals had an increasein plasma glucose levels produced by thestress. This effect was significantlygreater in the obese animals than in theirlean littermates (Fig. 2). Similarly, eventhough plasma insulin decreased in allanimals after stress, the decrease was sig-nificantly greater in obese animals. In alater study, this same group found thatstress-induced hyperglycemia easilycould be conditioned classically inC57BL/6J ob/ob mice, but not in a non-diabetic control strain (62). Taken to-gether, these data argue that the expres-sion of hyperglycemia in this geneticmodel of type II diabetes is dependent onexposure to stressful stimuli.

DIABETES CARE, VOLUME 15, NUMBER 10, OCTOBER 1992 1417

Stress and diabetes

5 soo

| | LEAN

PSS1 OBESE

C57BL/6J

. Q Normal

r-TSAL I 3 S

Figure 3—Effects of increasing dose of EPl in

plasma glucose in C57BL/6J ob/ob mice and their

lean littermates. SAL, saline-injected controls.

Dose is expressed as \ig salt/10 g body wt. Re-

sults are means ± SE glucose (mg/dl). From

Kuhn et al. (63). © by Physiology Biochemistry

and Behavior.

The mechanism by which stressappears to disregulate glucose metabo-lism in obese mice was studied by Surwitet al. (61). In one study, Surwit's groupstudied the effects of EPl on plasma glu-cose and insulin in both lean and obeseanimals. The experimental group re-ceived an injection of EPl bitartrate, andcontrol animals were injected with sa-line. Blood samples were drawn 1 h afterinjection. The results indicated that theeffects of EPl were analogous to those ofstress. That is, EPl produced an increasein plasma glucose in all animals, withobese mice showing a greater responsethan their lean littermates. Likewise, EPldecreased plasma insulin only in obesemice. In a follow-up study, Kuhn et al.(63) investigated the contribution of al-tered peripheral sympathetic function tothe exaggerated glycemic response ofob/ob mice to stress. Obese mice andtheir lean littermates were injected withone of three doses of EPl bitartrate orphentolamine mesylate. EPl was admin-istered in the dose range from 1-5 ng/10g body weight, whereas plasma glucoseresponses were maximal at the lowestdose of EPl tested in the obese mice,suggesting that the dose-response rela-tionship is altered in obese animals (Fig.3). Moreover, phentolamine, an a-ad-

Control EPl

Figure 4—Effects of diet-induced obesity and

strain on serum glucose response to EPl in

C57BL/6J (diabetes-prone) and A/] (diabetes-

resistant) mice. EPl was injected subcutaneous^

(0.5 \ig/10 g body wt). Control mice received

equal volume injections of saline. Results are

means ± SE. From Surwit et al. (65). © by the

American Diabetes Association.

renergic antagonist, produced a greaterincrease in plasma insulin in ob/ob micethan the lean littermates. This suggeststhat increased sensitivity to catechol-amines may be largely a-adrenergic. Al-tered peripheral responses to sympa-thetic stimuli are therefore important inthe exaggerated glycemic responses ofob/ob mice to stress and may be an etio-logical factor in the development of dia-betes in these animals. Fujimoto et al.(64) observed a similar exaggerated sen-sitivity to EPl in the KK mouse, anotheranimal model of type II diabetes. Fur-thermore, they observed that a-adrener-gic blockade with phentolamine couldprovoke an exaggerated insulin responsein KK mice compared with control mice,suggesting that KK, like ob/ob mice, havealtered adrenergic sensitivity.

Surwit et al. (65) also have shownthat altered adrenergic sensitivity mightbe a biological marker for the develop-ment of type II diabetes in the back-ground strain for the ob/ob mutation.They demonstrated that lean, nondia-betic C57BL/6J mice also show exagger-ated glucose response to EPl comparedwith several other strains (Fig. 4). Fur-thermore, they showed that these micedevelop type II diabetes when allowed tobecome obese on a high-fat, high-simple-

carbohydrate diet. That research teamthus speculated that altered sensitivity toadrenergic stimulation in the pancreas,liver, and possibly other sites may berelated to the pathophysiology of type IIdiabetes (50).

Finally, exaggerated glycemic re-activity to behavioral stress also appearsto be characteristic of at least some indi-viduals who are predisposed to develop-ing type II diabetes. Pima Indians are athigh risk for developing type II diabetes.Approximately 60% of Pima Indianseventually develop type II diabetes inadulthood, compared with 5% of thewhite population. In a previous study(66), young, euglycemic Pima Indiansshowed a disturbed glycemic response tobehavioral stress compared with whites.Pima Indians and whites were given amixed meal, which was followed 2 hlater by a 10-min mental-arithmeticstressor. Although both groups showedalmost identical normal glucose toler-ance in response to the meal (2 h pcglucose, 5.6 mM), 10 of 13 Pima Indiansshowed a small rise in glucose after amental-arithmetic stressor, whereas 7 of8 white control subjects did not (differ-ence in blood glucose between Pima In-dians and white control subjects afterstress was 1 mM). In that both groupsshowed similar cardiovascular and neu-roendocrine responses to the stress, itappears as though the diabetes-pronePima Indians have a specific glucoregu-latory defect that becomes apparent dur-ing such stimulation. As a marker for thedevelopment of type II diabetes, the di-rection of glycemic response to this stres-sor has a sensitivity of 76% and a spec-ificity of 87%. In that only —60% of ourPima Indian sample will develop type IIdiabetes, the sensitivity of the test may bemuch higher. These findings suggest thatenvironmental stress, which activates thesympathetic nervous system, may be par-ticularly deleterious to patients with typeII diabetes, and that methods to reducethe effects of stress may have some clin-ical utility in this disease (50).

No controlled studies have been

1418 DIABETES CARE, VOLUME 15, NUMBER 10, OCTOBER 1992

Surwit, Schneider, and Feinglos

conducted on how stress might affect theinitial onset of type II diabetes in hu-mans. However, as noted previously, theearliest recognition of how stress seem-ingly interacts with this disease was fromanecdotal reports of diabetes being pre-cipitated by acute life stress (1,2,3). Os-ier (3), in particular, appeared to differ-entiate type I from type II diabetes, inthat the latter form of the disease wasparticularly responsive to stress. Sincethe first awareness of the relationshipbetween stress and diabetes, numerousphysiological mechanisms have beenpostulated to explain how stress disruptsglucose metabolism in diabetes. One lineof research has focused on how glucoseintolerance can be induced by stress hor-mones or by stress itself in nondiabeticindividuals. Hamburg et al. (67) gavenormal volunteers a low-dose EPI infu-sion that produced elevations in plasmaEPI similar to those associated with thestress of a minor viral illness. They dem-onstrated that under these conditions,the deterioration in glucose tolerancewas significant. Using a different type ofexperimental paradigm, Wing et al. (68)investigated the effect of a standardizedpsychological stressor on the use of acarbohydrate load in normal individuals.Subjects were fed a carbohydrate load,and then were exposed to a stressor thatinvolved participation in competitivetasks. They found that an acute psycho-logical stressor delayed the peak glucoseresponse, and, therefore, apparently al-tered the ability of the subjects to absorbcarbohydrates. In a subsequent study,this group demonstrated that this alteredmetabolic response was attributable to astress-induced delay in gastrointestinaltransit time (69).

STRESS AND GLUCOSE CONTROLIN TYPE II DIABETES— A modestamount of literature on the effects ofstress on control of type II diabetes hasaccumulated over the past 15 yr. Grant etal. (70) looked at the role of stress interms of the relationship between actuallife events and fluctuations in the course

of diabetes mellitus. Subjects included15 patients with type I diabetes and 22patients with type II diabetes. Life eventswere measured with a modified versionof the Schedule of Recent Events (71).Their results suggested a relationship be-tween life events, particularly those of anegative nature, and changes in diabeticsymptomatology. Based on their results,they suggested that there may be life-event-responsive diabetic patients. How-ever, no attempt was made to differenti-ate stress responding in type I and type IIdiabetic patients. McClesky et al. (26)investigated the effect of a physical stres-sor, namely surgery and anesthesia, onglucagon levels in both normal, nondia-betic subjects and diabetic patients—allof whom were undergoing elective sur-gery. Serum glucagon and glucose levelswere repeatedly sampled during the pre-,intra-, and postoperative periods.Throughout the sampling period, the di-abetic patients showed a clear hypergly-cemic response to the surgical stress.

Although the mechanism of thisstress responsivity in diabetic patientshas not been studied directly, some evi-dence has accumulated for the alteredadrenergic sensitivity and responsivity inpatients with type II diabetes, and inanimal models of the disease. Linde andDeckert (72) and Robertson et al. (73)observed that a-adrenergic blockadewith phentolamine increased glucose-stimulated insulin secretion in type IIdiabetic patients; the latter groupshowed that this increase is 5 timesgreater in type II diabetic patients than innormal, nondiabetic control subjects.This suggests that a-adrenergic stimula-tion may have a greater effect on insulinrelease in diabetic patients than in nor-mal subjects. Most recently, Kashiwagi etal. (74) demonstrated that selectiveblockade of a-2 receptors increased bothinsulin secretion and glucose disposalrate after a mixed meal in type II diabeticpatients. Also, it has been reported thatglyburide, one of the sulfonylurea oralagents used to treat type II diabetes,binds to a-2 receptors in the pancreas,

suggesting that one effect of this drug oninsulin secretion may be caused by an-tagonism of adrenergic activity (75). Thevarious adrenergic effects on glucose me-tabolism are summarized in Fig. 1.

SUMMARY— For at least 300 years,physicians and patients have noted thatthe onset of diabetes often is preceded bysome significant life stress. This per-ceived relationship may simply be as-signable to the fact that two uncommonand salient life events often are perceivedas related more frequently than two morecommon events. However, some evi-dence does suggest that stress may pre-cipitate the onset of the disease or com-promise glycemic control once thedisease is established. Glucose toxicitythat results from chronic, intermittent,stress-induced elevations in blood glu-cose further may compromise pancreaticsecretory ability (12), leading to the pro-gression of the disease.

However, the literature on the ef-fects of stress in both experimental mod-els and in human type I diabetes is com-plicated and often contradictory, anddoes not allow us to reach a clear con-clusion about how stress impacts on thedisease. In animal models, the effects ofstress appear to depend on the type ofstress and the particular model studied.Also, because animal models of type Idiabetes rarely are treated with insulin,the relevance of these studies to clinicalproblems is not immediately apparent.Human studies are confusing, with somestudies showing hyperglycemic effects,some showing no effects, and someshowing idiosyncratic effects, with pa-tients demonstrating either hyper- or hy-poglycemia in response to the samestressor. These inconsistencies may re-flect true individual differences in meta-bolic response to stress that could berelated to other behavioral variables (e.g.,type A personality). On the other hand,confounding conditions, such as auto-nomic neuropathy, which develops overtime and would compromise any sympa-thetic nervous system response to stress,

DIABETES CARE, VOLUME 15, NUMBER 10, OCTOBER 1992 1419

Stress and diabetes

also may contribute to these apparentidiosyncratic responses to stress on thepart of different patients. At the moment,we can draw only limited conclusionsabout the effects of stress on type I dia-betes control or how these effects wouldbest be treated. Clearly, more research isneeded to determine the role of stress intype I diabetes. Further studies shouldcarefully exclude patients with symp-toms of autonomic neuropathy andshould test patients on multiple occa-sions to detect idiosyncratic patterns.

The potential role of stress intype II diabetes was noted a century agoby William Osier (3), and stress-reduc-ing maneuvers were recommended astreatment for this form of the disease.More recently, a theoretical rationale forthe importance of stress effects on glyce-mic control in type II diabetes hasemerged from animal studies. These uni-formly suggest that stress can affect gly-cemic control adversely in type II diabe-tes. Evidence from both animals andhumans suggests that individuals withtype II diabetes have altered adrenergicsensitivity in the pancreas, and perhapsother sites as well, which could makethem particularly sensitive to stressfulenvironmental stimulation. Other stimuliof sympathetic activity, such as dietaryfat and simple carbohydrates, also maycontribute more to the development ofdiabetes through this adrenergic mecha-nism. But, although substantial data ex-ists to show the theoretical importance ofstress in type II diabetes, no direct evi-dence demonstrates that stress plays aclinically significant role in the expres-sion or control of the human disease. Asin type I diabetes, clinical studies rarelyhave attempted to assess the role of in-dividual behavioral differences in identi-fying those patients who show stress hy-perglycemia or who may be particularlyresponsive to stress management. Fewstudies have followed patients longenough for generalizations on clinicaloutcome to be made. More clinical re-search is needed to determine the degree

to which stress affects the onset andcourse of diabetes.

Acknowledgments— Preparation of thismanuscript was supported, in part, by Grant5K02 MH 00303 from the National Instituteof Mental Health and Grants 1R01 DK 42923and 1R01 DK 43106 from the National Insti-tute of Diabetes, Digestive and Kidney Dis-eases to R.S.S.

References

1. Willis T: Pharmaceutice rationalis: Orthe exercitation of the operation of med-icines in humane bodies. In The Works ofThomas Willis. London, Dring, Harper, &Leigh, 1679

2. Maudsley H: The Pathology of Mind. NewYork, Appleton, 1899

3. Osier W: The Principles and Practice ofMedicine. New York, Appleton, 1892

4. Cannon WB: Bodily Changes in Pain, Hun-ger, Fear and Rage. New York, Mac-Millan, 1941

5. Surwit RS, Feinglos MN, Scovern AW:Diabetes and behavior: a paradigm forhealth psychology. Am Psychol 38:255-62, 1983

6. Surwit RS, Scovem AW, Feinglos MN:The role of behavior in diabetes care.Diabetes Care 5:337-42, 1982

7. Clutter WE, Bier DM, Shah SD, CryerPE: Epinephrine plasma metabolic clear-ance rates, and physiologic thresholdsfor metabolic and hemodynamic actionsin man. J Clin Invest 66:74-101, 1980

8. Williams RH, Porte D: The pancreas. InTextbook of Endocrinology. 5th ed.Williams RH, Ed. Philadelphia, PA,Saunders, 1974, p. 502-626

9. Landsberg L, Young JB: Sympathoadre-nal system: the regulation of metabolism.In Contemporary Endocrinology. Vol. 2.Ingbar SH, Ed. New York, Plenum,1985, p. 217-46

10. Boyle PJ, Ligett SB, Shah SD, Cryer PE:Direct muscarinic cholinergic inhibitionof hepatic glucose production in hu-mans. J Clin Invest 82:445-49, 1988

11. Guillemin R: Hypothalamus, hormonesand physiological regulation. In ClaudeBernard and the Internal Environment: A

Memorial Symposium. Debbs-Robin E,Ed. New York, Dekker, 1978, p. 137-56

12. Leahy, JL: Natural history of P-cell dys-function in NIDDM. Diabetes Care 13:992-1010, 1990

13. Capponi R, Kawada ME, Varela C, Var-gas L: Diabetes mellitus by repeatedstress in rats bearing chemical diabetes.Horm Metab Res 12:411-12, 1980

14. Huang SW, Plaut SM, Taylor G, Ware-heim BA: Effect of stressful stimulationon the incidence of streptozotocin in-duced diabetes in mice. Psychosom Med43:431-37, 1981

15. Roudier M, Portha B, Picon L: Glucocor-ticoid-induced recovery from streptozo-cin diabetes in the adult rat. Diabetes29:201-205, 1980

16. Nakhooda AF, Like AA, Chappell CI,Wei CN, Marliss EB: The spontaneouslydiabetic Wistar rat (the "BB" rat): studiesprior to and during the development ofthe overt syndrome. Diabetologia 14:199-207, 1978

17. Carter WR, Herman J, Stokes K, Cox DJ:Promotion of diabetes onset by stress inBB rat. Diabetologia 30:674-75, 1987

18. Lehman CD, Rodin J, McEwen B, BrintonR: Impact of environmental stress on theexpression of insulin-dependent diabe-tes mellitus. Behav Neurosci 105:241-45,1991

19. Robinson N, Fuller JH: Role of life eventsand difficulties in the onset of diabetesmellitus. J Psychosom Res 29:583-91,1985

20. Slawson PF, Flynn WR, Kollar EJ: Psy-chological factors associated with the on-set of diabetes mellitus. JAMA 185:166-70, 1963

21. Stein SP, Charles E: Emotional factors injuvenile diabetes mellitus: a study ofearly life experience of adolescent diabet-ics. Am] Psych 128:700-704, 1971

22. Lee JH, Konorska M, McCarty R: Physi-ological responses to acute stress in al-loxan and streptozotocin diabetic rats.Physiol Behav 45:483-89, 1989

23. Bellush LL, Rowland NE: Stress and be-havior in streptozotocin diabetic rats:biochemical correlates of passive avoid-ance learning. Behav Neurosci 103:144-50, 1989

24. Rowland NE, Bellush LL: Diabetes mel-

1420 DIABETES CARE, VOLUME 15, NUMBER 10, OCTOBER 1992

Surwit, Schneider, and Feinglos

litus: stress, neurochemistry and behav-ior. Neurosci Biobehav Rev 13:199-206,1989

25. Travis L: Stress, hyperglycemia, and ke-tosis. In Diabetes Mellitus in Children and

Adolescents. Travis L, Brouhard BH,Schreiner BD, Eds. Philadelphia, PA, W.B. Saunders, 1987, p. 137-46

26. McLesky CH, Lewis SB, Woodruff RE:Glucagon levels during anesthesia andsurgery in normal and diabetic patients(abstract). Diabetes 27:492, 1978

27. Baker L, Kaye R, Haque N: Studies onmetabolic homeostasis in juvenile diabe-tes mellitus II: role of catecholamines.Diabetes 16:504-505, 1967

28. Hinkle LE, Evans FM, Wolf S: Studies indiabetes mellitus III: Life history of threepersons with labile diabetes, and the re-lation of significant experiences in theirlives to the onset and course of theirdisease. Psychosom Med 13:160-83,1951

29. Hinkle LE, Evans FM, Wolf S: Studies indiabetes mellitus IV: Life history of threepersons with relatively mild, stable dia-betes, and relation of significant experi-ences in their lives to the onset andcourse of the disease. Psychosom Med1:184-202, 1951

30. Hinkle LE, Wolf S: Importance of lifestress in course and management of dia-betes mellitus. JAMA 148:513-20, 1952

31. Vandenbergh RL, Sussman KE, Titus CC:Effects of hypnotically induced acuteemotional stress on carbohydrate andlipid metabolism in patients with diabe-tes mellitus. Psychosom Med 28:382-90,1966

32. Vandenbergh RL, Sussman KE, VaughanGD: Effects of combined physical antici-patory stress on carbohydrate-lipid me-tabolism in patients with diabetes melli-tus. Psychosomatics 8:16-19, 1967

33. Chase HP, Jackson GG: Stress and sugarcontrol in children with insulin-depen-dent diabetes mellitus. J Pediatr 98:1011-13, 1981

34. Brand AH, Johnson JH, Johnson SB: Lifestress and diabetic control in childrenand adolescents with insulin-dependentdiabetes. J Pediatr Psychol 11:481-95,1986

35. Kemmer F, Bisping R, Steingruber H,

Baar H, Hardtmann R, Schlaghecke R,Berger M: Psychological stress and met-abolic control in patients with type I di-abetes mellitus. N Engl J Med 314:1078-84, 1984

36. Gilbert BO, Johnson SB, Silverstein J,Malone J: Psychological and physiologi-cal responses to acute laboratory stres-sors in insulin-dependent diabetes mel-litus adolescents and nondiabeticcontrols. J Pediatr Psychol 14:577-91,1989

37. Bradley C: Psychophysiological aspectsof the management of diabetes mellitus.MJ Mental Health 11:117-32, 1982

38. Carter WR, Gonder-Frederick LA, CoxDJ, Clarke WL, Scott D: Effect of stresson blood glucose in IDDM (Letter). Dia-betes Care 8:411-12, 1985

39. Stabler B, Surwit RS, Lane JD, MorrisMA, Litton J, Feinglos MN: Type A be-havior pattern and blood glucose controlin diabetic children. Psychosom Med 49:313-16, 1987

40. Stabler B, Lane JD, Ross SL, Morris MA,Litton J, Surwit RS: Type A behavior pat-tern and chronic glycemic control in in-dividuals with IDDM. Diabetes Care 11:361-62, 1988

41. Femqvist E, Gunnarsson R, Linde B: In-fluence of circulating epinephrine on ab-sorption of subcutaneously injected in-sulin. Diabetes 37:694-701, 1988

42. Sherwin RS, Shamoon H, Jacob R, SaccaL: Role of counterregulatory hormonesin the metabolic response to stress innormal and diabetic humans. In RecentAdvances in Obesity and Diabetes Research.

Melchionda N, Horwitz DL, Schade DS,Eds., New York, Raven, 1984, p. 327-44.

43. Cryer PE: Decreased sympathochromaf-fin activity in IDDM. Diabetes 38:405-409, 1989

44. Locatelli A, Franzetti I, Lepore G, MaglioML, Gaudio E, Caviezel F, Pozza G: Men-tal arithmatic stress as a test for evalua-tion of diabetic sympathetic autonomicneuropathy. Diabetic Med 6:490-95,1989

45. Davidson MB: Diabetes Mellitus: Diagnosis

and Treatment. New York, Wiley, 198146. Bell GI: Molecular defects in diabetes

mellitus. Diabetes 40:413-22, 1991

47. Giugliano D: Morphine, opioid peptides,and pancreatic islet function. DiabetesCare 7:92-98, 1984

48. Feldberg W, Pike DA, Stubbs WA: Onthe origin of non-insulin-dependent di-abetes, Lancet 1:1263-64, 1985

49. Ipp E, Dhorajiwala J, Pugh W, MoossaAR, Rubinstein AH: Effects of an en-kephalin analog on pancreatic endocrinefunction and glucose homeostasis in nor-mal and diabetic dogs. Endocrinology111:461-63, 1982

50. Surwit RS, Feinglos MN: Stress and au-tonomic nervous system in type II diabe-tes: a hypothesis. Diabetes Care 11:83—85, 1988

51. Van Loon GR, Appel N: Endorphin-induced hyperglycemia is mediated byincreased central sympathetic outflow toadrenal medulla. Brain Res 204:236-41,1981

52. Lori CF, Lori GT, Buchwald KW: Themechanism of epinephrine action vs.changes in blood sugar, lactic acid andblood pressure during continuous intra-venous injection of epinephrine. Am JPhysiol 93:273-83, 1930

53. Woods SC, Smith PH, Porte D: The roleof the nervous system in metabolic reg-ulation and its effects on diabetes andobesity. In Handbook of Diabetes Mellitus.Vol. 3. Brownless M, Ed. New York, Gar-land, 1981, p. 208-71

54. Berne C, Fagius J, Niklasson F: Sympa-thetic response to oral carbohydrate ad-ministration. J Clin Invest 84:1403-409,1989

55. Kaufman LN, Peterson MM, Smith SM:Hypertension and sympathetic hyperac-tivity induced in rats by high-fat or glu-cose diets. Am J Physiol 260:E95-100,1991

56. Dulin WE, Gerritsen GC, Chang A: Ex-perimental and spontaneous diabetes inanimals. In Diabetes Mellitus: Theory andPractice. 3rd ed. Ellenberg M, Rifkin H,Eds. New York, Medical ExaminationPublishing Company, 1983, p. 361-408

57. Mikat EM, Hackel DB, Cruz FT, LebovitzHE: Lowered glucose tolerance in thesand rat (psammonys obesus) resultingfrom esophageal intubation. Proceedingsof the Society for Experimental Biology and

Medicine 139:1390-91, 1972

DIABETES CARE, VOLUME 15, NUMBER 10, OCTOBER 1992 1421

Stress and diabetes

58. Beloff-Chain A, Freund N, RookledgeKA: Blood glucose and serum insulinlevels in lean and genetically obese mice.Horm Metab Res 1:374-78, 1975

59. Grundleger ML, Godbole VY, ThenenSW: Age dependent development of in-sulin resistance of soleus muscle in ge-netically obese (ob/ob) mice. Am] Physiol239:363-71, 1980

60. Dubuc PO, Cahn PJ, Ristimaki S, WillisPL: Starvation and age effects on glyco-regulation and hormone levels ofC57BL/6J ob/ob mice. Horm Metab Res

14:532-35, 1982

61. Surwit RS, Feinglos MN, Livingston EG,Kuhn CM, McCubbin JA: Behavioral ma-nipulation of the diabetic phenotype inob/ob mice. Diabetes 33:616-18, 1984

62. Surwit RS, McCubbin JA, Kuhn CM, Mc-Gee D, Gerstenfeld DA, Feinglos MN:Alprazolam reduces stress hyperglycemiain ob/ob mice. Psychosom Med 4 8 : 2 7 8 -

82, 198663. Kuhn CM, Cochrane C, Feinglos MN,

Surwit RS: Exaggerated peripheral re-sponsivity to catecholamines contributesto stress-induced hyperglycemia in the

ob/ob mouse. Physiol Biochem Behav 26:

491-95, 198764. Fujimoto K, Sakaguchi T, Ui M: Adren-

ergic mechanisms in the hyperglycaemiaand hyperinsulinaemia of diabetic KKmice. Diabetologia 20:568-72, 1981

65. Surwit RS, Kuhn CM, Cochrane C, Mc-Cubbin J A, Feinglos, MN: Diet-inducedtype II diabetes in C57BL/6J mice. Dia-betes 37:1163-67, 1988

66. Surwit RS, McCubbin JA, Feinglos MN,Esposito-Del Peunte A, Lillioja S: Glyce-mic reactivity to stress: a biologic markerfor development of type 2 diabetes (Ab-stract). Diabetes 39 (Suppl. 1):8A, 1990

67. Hamburg D, Hendler R, Sherwin RS: In-fluence of small increments of epineph-rine on glucose tolerance in normal hu-mans. Ann Intern Med 93:566-68, 1980

68. Wing RR, Epstein LH, Blair E, NowalkMP: Psychologic stress and blood glucoselevels in nondiabetic subjects. PsychosomMed 47:558-64, 1985

69. Blair EH, Wing RR, Wald A: The effectsof laboratory stressors on glycemic con-trol and gastrointestinal transit time. Psy-

chosom Med 53:133-43, 199170. Grant I, Kyle GC, Teichman A, Mendels

J: Recent life events and diabetes inadults. Psychosom Med 36:121-28, 1974

71. Holmes TH, Rahe RH: The social read-justment rating scale. J Psychosom Res 11:213-18, 1967

72. Linde J, Deckert T: Increase of insulinconcentration in maturity-onset diabet-ics by phentolamine (Regitine) infusion.Horm Metab Res 5:391-95, 1973

73. Robertson PR, Halter JB, Porte D: A rolefor alpha-adrenergic receptors in abnor-mal insulin secretion in diabetes melli-tus. J Clin Invest 57:791-95, 1976

74. Kashiwagi A, Harano Y, Suzuki M, Ko-jima H, Harada M, Nisho Y, Shigeta Y:New a2-adrenergic blocker (DG-5128)improves insulin secretion and in vivoglucose disposal in NIDDM patients. Di-abetes 35:1085-89, 1986

75. Cherksey B, Altszuler N: Tolbutamideand glyburide differ in effectiveness todisplace a- and |3- adrenergic radioli-gands in pancreatic islet cells and mem-branes. Diabetes 33:499-503, 1984

1422 DIABETES CARE, VOLUME 15, NUMBER 10, OCTOBER 1992