doi:10.1152/ajpendo.00139.2003 287:E371-E385, 2004.Am J Physiol Endocrinol MetabAndrea Caumo and Livio LuziConsiderations on shape and functionFirst-phase insulin secretion: does it exist in real life?

You might find this additional info useful...

112 articles, 42 of which can be accessed free at:This article cites http://ajpendo.physiology.org/content/287/3/E371.full.html#ref-list-1

10 other HighWire hosted articles, the first 5 are:This article has been cited by

  [PDF] [Full Text] [Abstract]

, November 1, 2008; 88 (5): 1277-1283.Am J Clin NutrGowerPaul B Higgins, José R Férnández, W Timothy Garvey, Wesley M Granger and Barbara AEuropean American childrenEntero-insular axis and postprandial insulin differences in African American and 

[PDF] [Full Text] [Abstract], March , 2010; 50 (3): 246-256.J Clin Pharmacol

Hanna E. Silber, Nicolas Frey and Mats O. KarlssonHealthy VolunteersAn Integrated Glucose-Insulin Model to Describe Oral Glucose Tolerance Test Data in 

[PDF] [Full Text] [Abstract], May , 2010; 298 (5): E1078-E1087.Am J Physiol Endocrinol Metab

Maude Le GallPatricia Serradas, Anne Lorsignol, Luc Pénicaud, Edith Brot-Laroche, Armelle Leturque and Emilie Stolarczyk, Christophe Guissard, Aurélien Michau, Patrick C. Even, Alexandra Grosfeld,intakeDetection of extracellular glucose by GLUT2 contributes to hypothalamic control of food 

[PDF] [Full Text] [Abstract], May , 2011; 300 (5): R1126-R1133.Am J Physiol Regul Integr Comp Physiol

Hui Min Chan, Ritesh Jain, Bo Ahrén, Giovanni Pacini and David Z. D'Argeniointravenous glucose administration in miceEffects of increasing doses of glucagon-like peptide-1 on insulin-releasing phases during 

[PDF] [Full Text], March , 2012; 61 (3): 562-563.Diabetes

Adrian Vella and Alan R. Zinsmeister-Cell FunctionβPredicting Diabetes Using Measures of

including high resolution figures, can be found at:Updated information and services http://ajpendo.physiology.org/content/287/3/E371.full.html

can be found at:AJP - Endocrinology and Metabolismabout Additional material and information http://www.the-aps.org/publications/ajpendo

This information is current as of April 23, 2012. 

http://www.the-aps.org/.20814-3991. Copyright © 2004 by the American Physiological Society. ISSN: 0193-1849, ESSN: 1522-1555. Visit our website atorganization. It is published 12 times a year (monthly) by the American Physiological Society, 9650 Rockville Pike, Bethesda MD

publishes results of original studies about endocrine and metabolic systems on any level ofAJP - Endocrinology and Metabolism

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

Minireview

First-phase insulin secretion: does it exist in real life?Considerations on shape and function

Andrea Caumo1 and Livio Luzi1,2

1Unit of Nutrition and Metabolism, Department of Medicine, San Raffaele Scientific Institute,20132 Milano; and 2Faculty of Exercise Sciences, University of Milano, 20100 Milano, Italy

Submitted 31 March 2003; accepted in final form 16 March 2004

Caumo, Andrea, and Livio Luzi. First-phase insulin secretion: does it exist inreal life? Considerations on shape and function. Am J Physiol Endocrinol Metab287: E371–E385, 2004; 10.1152/ajpendo.00139.2003.—To fulfill its preeminentfunction of regulating glucose metabolism, insulin secretion must not only bequantitatively appropriate but also have qualitative, dynamic properties that opti-mize insulin action on target tissues. This review focuses on the importance of thefirst-phase insulin secretion to glucose metabolism and attempts to illustrate therelationships between the first-phase insulin response to an intravenous glucosechallenge and the early insulin response following glucose ingestion. A clear-cutfirst phase occurs only when the �-cell is exposed to a rapidly changing glucosestimulus, like the one induced by a brisk intravenous glucose administration. Incontrast, peripheral insulin concentration following glucose ingestion does not bearany clear sign of biphasic shape. Coupling data from the literature with the resultsof a �-cell model simulation, a close relationship between the first-phase insulinresponse to intravenous glucose and the early insulin response to glucose ingestionemerges. It appears that the same ability of the �-cell to produce a pronounced firstphase in response to an intravenous glucose challenge can generate a rapidlyincreasing early phase in response to the blood glucose profile following glucoseingestion. This early insulin response to glucose is enhanced by the concomitantaction of incretins and neural responses to nutrient ingestion. Thus, under physio-logical circumstances, the key feature of the early insulin response seems to be theability to generate a rapidly increasing insulin profile. This notion is corroboratedby recent experimental evidence that the early insulin response, when assessed atthe portal level with a frequent sampling, displays a pulsatile nature. Thus, eventhough the classical first phase does not exist under physiological conditions, theoscillatory behavior identified at the portal level does serve the purpose of rapidlyexposing the liver to elevated insulin levels that, also in virtue of their up-and-downpattern, are particularly effective in restraining hepatic glucose production.

insulin pulsatile secretion; mathematical models; early-phase insulin secretion

IN THE PAST THREE DECADES, the relevance of insulin secretionabnormalities in the pathogenesis of type 2 diabetes mellitus havebeen extensively debated (32, 43, 86, 93, 97), and a consensus hasbeen reached that, to fulfill its pivotal role in regulating glucosemetabolism, insulin secretion must not only be quantitativelyappropriate, but also possess qualitative, dynamic features thatoptimize insulin action on target tissues. In particular, increasingemphasis has been placed on the importance of the so-calledfirst-phase insulin secretion to glucose homeostasis (27, 35, 55).The aim of this review is to address the following questions. Doesa first-phase insulin secretion really exist in in vivo physiology?What is the relationship between the first-phase insulin responseto a brisk intravenous glucose challenge and the early insulinresponse following glucose ingestion? What are the effects ofsuch insulin responses on glucose metabolism?

Before these issues are addressed, a precise definition offirst-phase insulin secretion is required. For the purpose of the

present review, five different modes (or phases) of insulinsecretion can be identified: 1) basal insulin secretion is the wayinsulin is released in the postabsorptive state; 2) the cephalicphase of insulin secretion is evoked by the sight, smell, andtaste of food (before any nutrient is absorbed by the gut) and ismediated by pancreatic innervation (2, 63); 3) first-phaseinsulin secretion is defined as the initial burst of insulin, whichis released in the first 5–10 min after the �-cell is exposed toa rapid increase in glucose (or other secretagogues); 4) after theacute response, there is a second-phase insulin secretion, whichrises more gradually and is directly related to the degree andduration of the stimulus; 5) finally, a third phase of insulinsecretion has been described, albeit only in vitro (49). Duringall these stages, insulin is secreted, like many other hormones,in a pulsatile fashion, resulting in oscillatory concentrations inperipheral blood. Oscillations include rapid pulses (recurring ev-ery 8–15 min) superimposed on slower, ultradian oscillations

Address for reprint requests and other correspondence: L. Luzi, San RaffaeleScientific Institute, Via Olgettina, 60, 20132 Milano, Italy (E-mail: luzi.livio@hsr.it).

The costs of publication of this article were defrayed in part by the paymentof page charges. The article must therefore be hereby marked “advertisement”in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Am J Physiol Endocrinol Metab 287: E371–E385, 2004;10.1152/ajpendo.00139.2003.

0193-1849/04 $5.00 Copyright © 2004 the American Physiological Societyhttp://www.ajpendo.org E371

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

(recurring every 80–120 min) that are closely related to fluctua-tions in glucose concentration (a comprehensive treatment ofinsulin pulsatility can be found in Refs. 64, 73, 88, and 89).

Insulin secretion can be induced by other energetic sub-strates besides glucose (particularly amino acids), as well as byhormones and drugs. In the following review, we will focus onthe insulin response to glucose. Of crucial importance to thepresent review is to make a clear distinction between theinsulin response following an intravenous glucose challengeand that following glucose or food ingestion. The intravenousadministration of glucose triggers a biphasic insulin response(featuring a rapid increase with a peak, an interpeak nadir, anda subsequent slower increasing phase) only when glucoseconcentration increases rapidly, as after a glucose bolus or aglucose infusion determining a square wave of hyperglycemia.A “slow-ramp” glucose input induces gradually larger secre-tion without a well-defined first phase (48, 107). More impor-tantly, a well-defined first phase is lacking under physiologicalconditions, i.e., when glucose is given orally. Thus, in thecourse of this review, we will refer to first-phase insulinsecretion only in the context of “rapid” intravenous glucosestimuli and will denote as early phase the insulin responseobserved during the initial 30 min that follow glucose ingestion(Fig. 1).

The review is organized as follows. First, we will brieflydescribe the methods used to quantify insulin secretion in vivo.Then, we will summarize investigations assessing first-phaseinsulin secretion in response to glucose stimuli in vitro and invivo and the most significant modeling attempts that have beenput forth to better understand the mechanisms underlyingbiphasic insulin secretion. We will then provide an essentialcompendium of the studies that investigated the impact offirst-phase insulin secretion on glucose metabolism. Subse-quently, we will tackle insulin secretion under physiologicalconditions and try to elucidate the relationships between theinitial insulin response to intravenous glucose (first phase) andoral glucose (early phase) through the use of a model simula-tion. After that, we will move on to summarize the beneficialeffects that the early insulin secretion exerts on prandial glu-cose metabolism. Finally, we will briefly highlight recentstudies showing that the early insulin response observed inperipheral blood after glucose ingestion is the dampened ver-

sion of a portal oscillatory pattern that repeatedly exposes theliver to a first-phase-like insulin pattern.

MEASUREMENT OF INSULIN SECRETION

In assessing insulin secretion in vivo, one faces severaldifficulties and must use mathematical models as an aid (asreviewed in Refs. 57, 71, and 114). Insulin is secreted by the�-cell into the portal vein, which is usually not accessible. Inaddition, plasma insulin levels reflect not only insulin secretionbut also insulin distribution and degradation (i.e., the kinetics).Finally, before diluting into the peripheral circulation, insulinpasses through the liver, where it undergoes a large (�50%)and probably time-varying extraction (45). The latter short-coming can be bypassed by exploiting the properties of C-peptide. C-peptide is cosecreted with insulin on an equimolarbasis but, unlike insulin, is not extracted by the liver (Fig. 2).The gold standard to derive C-peptide secretion from C-peptideconcentrations is based on C-peptide kinetic modeling anddeconvolution [the so-called Eaton-Polonsky approach (37,87)]. Deconvolution is a procedure that allows one to infer the

Fig. 1. Differences between the incremental(above basal) plasma insulin concentration pro-files in response either to an intravenous glucoseinfusion producing a square wave of hypergly-cemia (A) or to a meal (B). The response to theintravenous glucose challenge shows a biphasicprofile, with distinct 1st- (0–10 min) and 2nd-phase (10–30 min) insulin-secretory responses.In contrast, the insulin response to the physio-logical entry of glucose does not exhibit a clearlydistinguishable biphasic pattern. Of note is thatthe insulin response observed after food inges-tion cannot be accounted for solely by the asso-ciated changes in the blood glucose level but alsodepends on other factors, such as the presence offree fatty acids and other secretagogues in themeal, the neurally activated cephalic phase, andgastrointestinal hormones.

Fig. 2. Schematic representation of the secretion and kinetics of insulin andC-peptide. The �-cell releases equimolar amounts of insulin (top) and C-peptide (bottom). Part of released insulin is degraded by the liver, but all of thereleased C-peptide reaches the periphery, where both peptides are measured inplasma. Both insulin and C-peptide follow 2-compartment kinetics, although1-compartment descriptions are frequently used in the literature (especially forinsulin, as discussed in Ref. 44). This general schema is at the basis of variousmodeling approaches to the estimation of insulin secretion in vivo.

Minireview

E372 FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

secretion rate of a substance from its peripheral concentrationsby mathematically removing the impact of the substance ki-netics. To estimate C-peptide secretion by deconvolution,knowledge of C-peptide kinetics is thus required. C-peptidekinetics are usually determined in a separate experiment byanalyzing the C-peptide decay curve following a bolus injec-tion of biosynthetic C-peptide. If C-peptide kinetics cannot beobtained in a separate experiment, the population-based C-peptide kinetic parameters reported by Van Cauter et al. (118)can be used as a reliable alternative for all those situationswhere renal function, and thus C-peptide kinetics, are notimpaired. Deconvolution has been extensively used to recon-struct insulin secretion in various experimental conditions (14,37, 56, 87, 103). Recent advances in the deconvolution tech-niques (85, 108) allow us to cope better with the highlydynamic nature of the first-phase insulin response.

A simpler alternative to deconvolution, denoted the “com-bined model,” has been introduced by Volund et al. (119) andsubsequently applied and refined by Watanabe and colleagues(121, 122). The combined model allows for the estimation ofinsulin secretion without separate experimental assessment or apriori assumption of the C-peptide kinetic parameters. This isaccomplished by simultaneously modeling C-peptide and in-sulin kinetics and exploiting the fact that C-peptide and insulindata are two sources of information concerning the samesecretion rate (as shown in Fig. 2). The combined model hasbeen validated against deconvolution under experimental con-ditions mimicking the slow changes in insulin secretion ob-served during an oral glucose tolerance test (OGTT) (62).Another approach to measure insulin secretion is the minimalmodel of Toffolo et al. (115) (“minimal” denotes here anoptimal degree of complexity). The minimal model uses onlyC-peptide data and individualizes C-peptide kinetics using theparameters of Van Cauter et al. (118). Its distinctive andappealing feature is an explicit description of glucose controlon �-cell secretion. This allows the model to provide not onlyinsulin secretion but also indexes of �-cell responsivity toglucose that can be helpful for the characterization of impairedmechanisms of insulin secretion in pathophysiological states.The model was originally developed to interpret the intrave-nous glucose tolerance test (IVGTT) (115), but it has beenrecently adapted to other experimental conditions such asup-and-down glucose staircase profiles (113) and the OGTT(13). The model has been validated against deconvolution inexperimental protocols characterized by slow changes in insu-lin secretion (e.g., up-and-down staircase profiles) (14). Other�-cell models that attempt to describe the functional depen-dence of insulin secretion on glucose concentration have beendeveloped in recent years, especially to quantify insulin secre-tion during experimental protocols reproducing normal lifeconditions (e.g., mixed-meal tests and OGTTs) (29, 56, 72).These models, which differ for the specific assumptions aboutthe control exerted by glucose on �-cell secretion, have beenrecently reviewed (81).

Although the quantitative assessment of insulin secretion isusually based on the measurement of C-peptide, it must berecognized that, from a qualitative standpoint, the profile ofplasma insulin concentration is better than that of C-peptide inreproducing the shape of insulin secretion (104). This occursbecause the half-life of insulin (�10 min) (44) is markedlyshorter than that of C-peptide (�35 min) (118). Thus alterna-

tive approaches, which simply rely on insulin concentrations,to estimating surrogate indexes of first- and second-phaseinsulin secretion, have been developed. For instance, insulinsecretion indexes can be derived from a hyperglycemic clampby calculating areas under the insulin concentration curve(from 0 to 10 min for the first phase, and from 10 min onwardfor the second phase). Another widely used index of first-phaseinsulin secretion is the acute insulin-secretory response toglucose (AIRGlc), which is calculated as the mean incrementalincrease in plasma insulin concentration in samples obtained at3, 4, and 5 min of an IVGTT (25). A more sophisticatedapproach to the estimation of indexes of �-cell function fromIVGTT insulin data is based on the minimal-model descriptionof the functional relationship between the plasma glucoseconcentration and the posthepatic insulin delivery rate (112).

In studying the insulin response to an oral glucose load, firstand second phases are not clearly distinguishable. During ameal, a typical index of early insulin response is the insulinincrement at 30 min (�I30). When this index is normalized tothe corresponding glucose increment (�I30/�Glc30), it takes thename insulinogenic index and measures a sort of sensitivity ofthe �-cell to glucose (84).

BIPHASIC NATURE OF THE INSULIN RESPONSE: IN VITROAND IN VIVO STUDIES

The presence of a biphasic dynamics of �-cell release inresponse to glucose has been extensively investigated in vitro,both at the level of the islets comprising the endocrine pancreasand at the organ level. The physiology of insulin secretion atthe islet level involves several molecular events whose descrip-tion goes beyond the scope of this review. We refer the readerinterested in having an up-to-date overview of insulin secretionat the islet level to a recent collection of review articlesappearing in a monographic issue of the journal Diabetesdedicated to insulin secretion (Diabetes 51, Suppl 1, 2002).Insulin secretion at the organ level was extensively studied byGrodsky and colleagues (an overview of their findings can befound in Ref. 50). They used the isolated rat pancreas perfusedwith a buffer containing glucose to study insulin secretiondynamics in response to a variety of glucose stimuli, such asstepwise and staircase increments as well as short pulses (30,48, 51, 52). These studies showed that the �-cell responds to aprolonged single step of glucose with a rapid and transient firstphase of insulin secretion followed by a slower but increasing-in-time second phase (Fig. 3). The magnitude of the first phasein response to a single-step glucose stimulus increases withincreasing doses of glucose. Specifically, the amount of insulinreleased during the first phase is a sigmoidal function ofglucose concentration (with a half-maximum for glucose of�135 mg/dl). When the glucose stimulus is repeated in shortintervals, the first-phase insulin response is inhibited; in con-trast, if longer time intervals are used, enhancement of thefirst-phase insulin response is observed at the second stimula-tion. Thus the magnitude of the first phase is dependent on thepancreas’s past history of stimulation by glucose. The firstphase takes place almost instantaneously after the beginning ofthe stimulation, and, at any glucose level, the rapid rise anddecay of the first phase occur within 3–4 min. The secondphase can last several hours if the �-cell is continuouslyexposed to glucose. A desensitization phase, characterized by

Minireview

E373FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

a sustained but lower insulin release, can be demonstrated forup to 24 h with prolonged exposure to glucose. The detailedportrait of insulin secretion provided by Grodsky and cowork-ers using the perfused rat pancreas was corroborated by Berg-man and Urquhart (9) using a peculiar experimental approach.These investigators devised a method, called the “pilot gland”approach, to perfuse the pancreas with whole blood rather thanan artificial medium, thus exposing the organ to the metabolicfactors that are normally present in vivo. With this approach,the pancreas extirpated from a small dog (i.e., the pilot gland)is perfused with the arterial blood of an intact, anesthetized,large dog. The weight ratio between the two dogs (1:7) is suchthat dynamic tests can be performed on the pilot pancreaswithout causing evident metabolic alterations in the large dog.

Phasic insulin secretion can be demonstrated in vivo byadministering glucose intravenously in such way as to quicklyincrease plasma glucose concentration. Typical glucose stimuliare the bolus injection (whose clinical counterpart is theIVGTT), the primed constant infusion, and the primed variableinfusion establishing a square wave of hyperglycemia [whoseclinical counterpart is the hyperglycemic clamp (33)]. Suchinputs expose the �-cell to an abrupt rise in glucose concen-tration and are able to elicit a biphasic insulin response (23, 24,33, 65, 94). Let us focus our attention on the advantages anddrawbacks of the IVGTT compared with the hyperglycemicclamp. The IVGTT is the protocol of choice when simplicity isrequired, because the hyperglycemic clamp (whose featuresand applications have been reviewed in Ref. 40) entails asophisticated experimental procedure, is labor intensive, andrequires considerable expertise. On the other hand, one advan-tage of the hyperglycemic clamp over the IVGTT is that thepancreas is exposed to the same glucose increment for thedesired duration, whereas during the IVGTT the glucose peakmay differ among subjects and glucose concentration decays ata rate that depends on the individual’s glucose tolerance. Inaddition, the prolonged elevation of glucose concentration that

the hyperglycemic clamp brings about allows a clearer deter-mination of the second-phase insulin secretion with respect tothe IVGTT. For these reasons, the hyperglycemic clamp isconsidered the gold standard method for the assessment of thebiphasic insulin response in vivo. How fast does the pancreasrespond to an intravenous glucose challenge? When blood canbe sampled from the portal vein, the time lag between the riseof glucose concentration to the peak insulin response is short:1–2 min. When only peripheral sampling is available, thebiphasic insulin response is delayed and dampened due to thepassage through the liver and distribution in the peripheralblood. For instance, during a hyperglycemic clamp (Fig. 4), thefirst phase occurs between 3 and 5 min; then the plasma insulinfalls rapidly, reaching a nadir at 10 min; finally, a second phaseis observed during which plasma insulin increases almostlinearly as long as the glucose stimulus is maintained. Theimpairment of the first-phase insulin response to intravenousglucose is a marker of the pathology of the �-cell. For example,the first-phase insulin response is almost invariably reduced in

Fig. 4. Hyperglycemic clamp data from our laboratory (means � SE in 7normal subjects). The glucose level in peripheral blood (A) increases rapidly inresponse to the exogenous glucose infusion. The insulin and C-peptide con-centrations (B) show a biphasic pattern, with a peak at 3–5 min, a nadir at 10min, and a slowly increasing time course thereafter. Of note is that the initialpeak of C-peptide is less pronounced than that of insulin because the half-lifeof C-peptide is longer than that of insulin (35 vs. 10 min).

Fig. 3. Insulin secretion from in vitro perfused pancreas of the rat in responseto steps in glucose concentration. Reproduced from O’Connor et al. (80) withthe permission of the American Journal of Physiology.

Minireview

E374 FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

patients with impaired glucose tolerance (IGT) or in the earlystages of type 2 diabetes, as shown in Refs. 24, 39, and 101 andreported in countless review articles (see, for instance, Refs. 32and 120).

EXPLANATION OF THE BIPHASIC INSULIN RESPONSE: ROLEOF MODELS

An abundance of research has been performed to shed lighton the mechanisms underlying the biphasic insulin responseobserved in an isolated and perfused pancreas or in an intactorganism following abrupt changes in glucose concentration.Many studies have investigated the macrophysiology of �-cellsecretion, whereas many others have tried to identify the basicbiochemical and molecular processes that are involved. In thefollowing, we will briefly deal with the former type of studies,and we refer the reader interested in having an up-to-dateoverview of insulin secretion at the molecular level to theaforementioned monographic issue of the journal Diabetesdevoted to insulin secretion (Diabetes 51, Suppl 1, 2002). Inthe studies that have addressed the biphasic insulin secretion interms of the macrophysiology of the event, mathematicalmodeling has played a relevant role. Two distinct modelingapproaches have been developed over the years: the “storage-limited model” and the “signal-limited model” (comparisons ofthese two model categories can be found in Refs. 78 and 80).

The storage-limited model was developed by Grodsky andcoworkers (48, 52) on the basis of experimental data of insulinsecretion obtained in vitro from the isolated perfused pancreas[it should be noted that secretion data obtained in vivo inhumans are compatible with this kind of model (94)]. Thestorage-limited model assumes that the �-cell contains twodistinct pools of insulin granules: a small (2%), labile poolaccessible for immediate release [which has been anatomicallyidentified with a pool of docking granules in the �-cell (31)]and a larger (98%), stable pool that feeds slowly into the labilepool. Glucose promotes the provision of new insulin and thetransfer from the storage to the labile pool. According to thismodel, the transient nature of the first-phase insulin release isthe consequence of the voiding of the labile pool early duringglucose stimulation, whereas the exhaustion corresponds to thenadir between the two phases of insulin secretion. Whensufficient time is allotted, the labile compartment is refilled bythe transfer of granules from the large insulin pool. Therefilling process is driven by glucose through a potentiatingsignal that accumulates gradually. With time, the potentiatingsignal causes the insulin content in the labile compartment torise, thus producing the increasing second-phase secretion. Onedrawback of the original formulation of the storage-limitedmodel (52) is that a constant size of the labile pool is notcapable of accommodating for the observation that progres-sively higher concentrations of glucose produce first-phaseresponses of increasing amplitude. For this reason, the modelwas subsequently refined by incorporating the assumption thatthe labile pool is not homogeneous but contains populations of�-cells (packets) with different threshold sensitivities to glu-cose (48). These packets release their insulin content in anall-or-nothing fashion when their thresholds to glucose areexceeded (Fig. 5) [the theory of the threshold secretory mech-anism was formalized by Licko (67)].

The signal-limited model was developed in a series ofpublications by Cerasi and coworkers (20–22, 77). Theseinvestigators abandoned the pool model in favor of the ideathat biphasic insulin release is the result of the dynamicinteraction between stimulatory and inhibitory events initiatedby glucose, each of them having its own kinetics and dosedependence. In this model, it is assumed that high levels ofglucose rapidly trigger a signal for the initiation of insulinrelease. This signal is thought to be proportional to the incre-ment of glucose concentration above basal. In addition to thisinitiation, glucose enhances the sensitivity of the islet, a pro-cess denoted as potentiation, which increases the rate of insulinrelease triggered by the initiating action of glucose. Thispotentiation is assumed to be time varying and to have arelatively long half-life. Finally, it is postulated that glucoseinduces an inhibitory effect on insulin release, thereby reduc-ing the stimulatory effects of initiation and potentiation. If thetime course of this inhibition is intermediary to those ofinitiation and potentiation, the superposition of these threesignals is able to produce the classical biphasic �-cell responsewith its distinctive features (Fig. 6, top). Based on this idea, aglucose-insulin model was constructed whose primary aim wasto measure the quantitative aspects (time characteristics, dosedependencies, etc.) of initiation, potentiation, and inhibitionfrom the analysis of individual stimulus-response experiments(22) (Fig. 6, bottom).

EFFECTS OF FIRST-PHASE INSULIN SECRETION ONGLUCOSE METABOLISM

Insulin plays a pivotal role in glucose homeostasis by virtueof its ability to inhibit hepatic glucose production and stimulateglucose uptake. The magnitude of these insulin effects dependson the concentration of insulin at the liver site and in theinterstitial fluids, respectively. Figure 7 shows the results of amodel simulation illustrating the stages that a biphasic insulinresponse goes through after being released by the �-cell. First,insulin passes through the liver, where it exerts some of itseffects; then, it enters the general circulation where is diluted;and finally, it is carried through the body and reaches theinterstitial fluids, where it interacts with the cell receptors (theinterstitial fluids are modeled as a compartment remote fromplasma, in accord with Refs. 7, 105, and 123).

Anatomically, the liver is the most obvious candidate forbeing the target organ for first-phase insulin secretion. Because

Fig. 5. Schematic diagram of the storage-limited (2-compartment) model ofinsulin secretion. The model embodies the hypothesis of a distribution ofinsulin packets that promptly release insulin when their thresholds to glucoseare exceeded. Continuous lines, fluxes of material; dashed lines, controlsignals. See text for explanation. Adapted from Grodsky (48) with the permis-sion of the Journal of Clinical Investigation.

Minireview

E375FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

insulin is secreted directly into the portal system, this ensuresa greater hepatic than peripheral insulinization (porto-systemicgradient). In addition, the shape of sinusoidal insulin closelyreproduces the shape of insulin secretion. Thus any directeffect that insulin is able to exert on the liver is magnified bythe presence of the transient but elevated insulin levels thatcharacterize the first phase. Insulin concentration in the inter-stitial fluids does not resemble insulin secretion as a result ofthe delay associated with the transit of insulin through theendothelial barrier. In particular, interstitial insulin bears notrace of the original biphasic shape. Nevertheless, the avail-ability of a large first-phase secretion has been shown topromote a quicker activation of glucose uptake in dogs (46).Thus it is likely that the first phase will have beneficial effectson virtually all the insulin-dependent processes that ensureglucose homeostasis. The overall result is an improvement ofglucose tolerance, as shown in humans by Calles-Escandonand Robbins (18). In that study, the authors used a briefintravenous infusion of somatostatin to selectively inhibit thefirst-phase insulin response to an IVGTT in normal subjects (inthis way, they acutely created a model to reproduce the loss ofthe first phase seen in the diabetic state). They found that theloss of the first-phase insulin response was associated with areduced glucose tolerance and a possible hampering of thethermic effects of ingested carbohydrates.

In the following paragraphs, we will summarize the resultsof studies investigating the benefits associated with first-phaseinsulin release. We will first illustrate the effects of the firstphase on peripheral tissues and then examine the effects of thefirst phase at the liver site.

Effects of first-phase insulin response on peripheral tissues.The time course of glucose uptake by peripheral tissues lagsconsiderably following insulin response. It has been suggestedthat this delay is due to the time it takes insulin to cross theendothelial barrier and reach the interstitial medium bathing

Fig. 6. Schematic description of the signal-limited model of insulin secretion.The model hinges on the hypothesis that insulin secretion is the result of thedynamic balance among proportional, potentiating, and inhibitory insulino-genic signals triggered by glucose. An example is given of the time course ofthese signals during a square-wave glucose stimulus (top) together with arepresentation of the model structure (bottom). Adapted from Nesher andCerasi (78) (Copyright © 2002 American Diabetes Association from Diabetes51: S53–S59, 2002. Reprinted with the permission of The American DiabetesAssociation).

Fig. 7. Model simulation of the incremental (above basal) insulin response to a square wave of hyperglycemia. The biphasicinsulin-secretory response (A) and the corresponding insulin concentration profiles in plasma (B) and in the interstitial fluids (C)are shown. It was assumed in the simulation that the steady-state ratio between plasma and interstitial insulin is 3:2 (123) and thatthe fractional clearance rate of the remote insulin compartment representing interstitial insulin is 0.045 min�1 (19). Of note is thatthe size of the 1st phase is progressively attenuated as insulin moves from the portal circulation to the systemic circulation and upto the interstitial fluids.

Minireview

E376 FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

the insulin-sensitive cells (10, 123). In particular, it has beenshown in dogs that changes in glucose uptake are closelypredicted by changes in hindlimb lymph insulin (representinginterstitial insulin bathing the skeletal muscle) (95). Thus, toappreciate the effect of the biphasic insulin response on glu-cose uptake, one should make reference to the time course ofinsulin in the interstitial fluids, not in plasma. As insulinreaches the interstitial fluids, the biphasic pattern is no longerappreciable, because the first phase is considerably attenuatedand merged with the second phase (Fig. 7). Even thoughinsulin loses its biphasic pattern when it reaches the interstitialfluids, a large first-phase insulin secretion causes remote insu-lin to increase more quickly than it otherwise would. Getty etal. (46) have shown in dogs that, when the contribution of thefirst-phase insulin response to the overall increase in insulin isof the order of �50%, this produces a rise in insulin lymphconcentration that is much more rapid than the one observedwhen only the second phase is present. This, in turn, promotesa quick activation of glucose uptake by the tissues.

Because first-phase insulin secretion determines a timecourse of interstitial insulin concentration that has beneficialeffects on glucose uptake, it is not unreasonable to hypothesizethat it may also promote a quick activation of insulin’s indirecteffect on hepatic glucose production. It has been establishedthat insulin’s inhibitory effect on hepatic glucose production isin part direct, i.e., due to the interaction between hepaticsinusoidal insulin with the hepatocyte, and in part indirect, i.e.,due to the action of insulin on extrahepatic tissues, the latteraction consisting mainly of the antilipolytic effect of insulin onadipose tissue (1, 26, 47, 66, 70, 96, 98). The insulin-induceddecrease of substrate flow to the liver and the subsequentinhibition of hepatic glucose production [i.e., the pillar of theso-called single-gateway hypothesis (6)] is a rather slow pro-cess. It has been suggested that the delay in insulin suppressionof lipolysis may be secondary to the delay in insulin reachingthe interstitial fluids. Thus it is likely that insulin concentrationin the interstitial fluids plays a crucial role also in regulatinginsulin’s indirect inhibition of hepatic glucose production.

Because the first-phase insulin response favors a more rapidelevation of insulin concentration in the interstitial fluids, asascertained by Getty et al. (46), it is plausible that insulin’sindirect inhibition of hepatic glucose production is also acti-vated more rapidly when the first-phase insulin response ispresent.

Effects of first-phase insulin response on the liver. Insulinhas a direct inhibitory effect on hepatic glucose production. Indog studies, insulin’s direct effect on hepatic glucose produc-tion has been found to be potent and more important thatinsulin’s indirect effect, at least in nondiabetic animals (26).Lewis et al. (66) have shown in humans that hepatic glucoseproduction is inhibited to a greater extent with portal comparedwith peripheral insulin delivery, with matched peripheral insu-lin levels. Maheux et al. (70) have shown in humans that anincrease in portal vein insulin concentration can rapidly inhibithepatic glucose production independently of an increase inperipheral plasma insulin concentration. Thus, by acutely ele-vating liver sinusoidal insulin concentration, first-phase insulinsecretion brings about a rapid inhibitory effect on hepaticglucose production. In addition, when the delayed version ofthe first phase reaches the interstitial fluids, the above-men-tioned indirect effects of insulin will also become manifest andcontribute to restrain hepatic glucose production. The relativeimportance and the precise timing of either the direct orindirect effects of first-phase insulin secretion on hepatic glu-cose production have not been fully elucidated yet, but exper-imental evidence has shown that the overall result on hepaticglucose production is remarkable and long lasting. In collab-oration with Dr. Ralph DeFronzo, one of us was the first todocument the crucial role played by first-phase insulin secre-tion in restraining hepatic glucose production (68). In thatstudy, normal young subjects underwent hyperglycemic clampstudies in which somatostatin was administered, basal gluca-gon and insulin levels were replaced, and exogenous insulinwas infused to mimic either a normal (first- and second-phase)or a second-phase-only insulin response (Fig. 8, A and C). Wefound that the presence of the first phase was associated with

Fig. 8. Experimental data (A and C) andglucose turnover results (B and D) of thestudy by Luzi and DeFronzo (68). In thatstudy, the hyperglycemic clamp was used inconjunction with the pancreatic clamp tech-nique (simultaneous basal glucagon and in-sulin replacement via exogenous infusion) todetermine the effects of the 1st-phase insulinresponse on glucose production and glucoseclearance. Two hyperglycemic clamp proto-cols were performed. In the 1st, an exoge-nous insulin infusion was used to mimic both1st- and 2nd-phase insulin response; in the2nd, only the 2nd-phase insulin response wasreproduced. The glucose and insulin concen-trations measured on the 2 occasions areshown in A and C, respectively. The glucoseturnover results (B and D) indicate that, in thepresence of the 1st-phase insulin response,glucose production is more rapidly and pro-foundly inhibited (B), whereas glucose clear-ance is not not significantly enhanced (D).Data are replotted from Luzi and DeFronzowith the permission of the American Journalof Physiology.

Minireview

E377FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

a more rapid and marked suppression of hepatic glucoseproduction (Fig. 8B) (a finding confirmed with a similar ex-perimental approach in Ref. 41), whereas glucose clearance(Fig. 8D) was not significantly affected (as previously shownin Ref. 12 with the euglycemic hyperinsulinemic clamp).Interestingly, despite the transient nature of the first phase, inour study the enhancement of the inhibition of hepatic glucoseproduction was still present 2 h after the beginning of the test.This suggests that priming the liver with insulin may producelong-term benefits, a notion that is confirmed under prandialconditions, as we will see in the following sections of thisreview.

Another way in which the first-phase insulin response canact on the liver is through the interaction with glucagon [see, inthis regard, the review by Unger (116)]. Steiner et al. (109)carried out experiments in dogs to investigate the ability of theindividual components of the biphasic insulin response tocounteract glucagon-stimulated hepatic glucose production.They infused glucagon intraportally to determine a fourfoldincrease in plasma glucagon and observed the effects onglucose metabolism under four conditions: 1) in the absence ofan incremental insulin release, 2) in the presence of a simulatedfirst-phase insulin release, 3) in the presence of a simulatedsecond-phase insulin release, and 4) in the presence of asimulated first- and second-phase insulin release. With thisexperimental approach, the investigators showed that bothphases of insulin release are important for full counterregula-tion of the action of glucagon on glucose metabolism. Whenthe first-phase insulin response is lacking, the insulin-to-glu-cagon ratio is rapidly altered in favor of glucagon, thus leadingto increased hepatic glucose production and hyperglycemia.Restoration of the first-phase insulin profile in plasma, even inthe absence of the second phase, curbs the glucagon-mediatedrise in plasma glucose concentration. In a subsequent study, thesame group of investigators showed that first-phase insulinresponse acts primarily by inhibiting hepatic glucose produc-tion through a reduction in gluconeogenesis (110).

INSULIN SECRETION UNDER PHYSIOLOGICAL CONDITIONS

We have seen that a clear-cut first-phase insulin responsecan be elicited in vivo resorting only to stimuli that rapidlyelevate blood glucose concentration. Such stimuli, however,are not physiological, because abrupt and marked increases inblood glucose concentration do not occur in nature. Underphysiological conditions, after a meal for instance, glucoseconcentration increases gradually, and the insulin responsemeasured in the peripheral blood does not bear any clear signof a biphasic shape.

What is the relationship between the biphasic insulin re-sponse following an intravenous glucose stimulus and the earlyinsulin response, much less biphasic, or not biphasic at all,observed under physiological conditions? In the attempt toanswer this question, it may be of some help to consider thebiphasic nature of insulin secretion from a teleological per-spective. A biphasic insulin response can be conceived as thesuperposition of a priming bolus and a secondary infusion ofinsulin, this combination allowing the rapid achievement ofelevated insulin concentrations. It is plausible that the same�-cell dynamic properties that are able to generate a biphasicsecretion in response to a brisk, intravenous glucose challenge

are still operating in response to the gradual entry of glucosefrom the gut. Thus, even though insulin secretion tends to loseits biphasic shape under physiological conditions, the earlyinsulin response may still be aimed to quickly shift glucosemetabolism from the fasting to the prandial state.

To gain some quantitative insight into this issue, we per-formed a simulation of glucose-stimulated �-cell secretionusing the minimal model of Toffolo et al. (113). This model(illustrated in Fig. 9) postulates that insulin secretion in re-sponse to glucose elevation is the sum of two components, onestatic and the other dynamic. The static component (governedby parameters � and �) is proportional to �glucose concen-tration (it should be noted that a delay between the glucosestimulation and the actual insulin release by the �-cell isassumed). The dynamic component (governed by parameterkd) is proportional to the derivative of glucose concentrationand accounts for the observation that rapid changes in glucoseconcentration enhance insulin secretion (25, 79). Coupling this�-cell model with a two-compartment model of insulin kinetics(whose parameters were taken from Ref. 87) and assuming a50% first-pass hepatic extraction of insulin, we were able topredict the �-cell secretion and the corresponding profile ofperipheral insulin concentration in response to two differentglucose stimuli. The two glucose stimuli were intravenousglucose infusions producing either a square wave of hypergly-cemia or a gradual glycemic elevation mimicking the postpran-dial hyperglycemia (in the following, the latter glucose inputwill be denoted meal-like study). The model operated underconditions reproducing either a normally functioning �-cell ora defective �-cell with a selective impairment of the first-phaseinsulin response. To reproduce a normally functioning �-cell,we fixed the parameters of the model (�, �, kd) to the valuesthat are found in normal subjects in response to intravenousglucose administration. To reproduce a �-cell with a defectivefirst phase, we left unchanged the parameters � and � of thestatic control but assumed a 66% reduction in parameter kd of

Fig. 9. Schematic representation of the model of glucose-stimulated �-cellsecretion developed by Toffolo et al. (113) and subsequently applied to the oralglucose tolerance test (OGTT) by Breda et al. (13). The model includes a staticcontrol, i.e., a secretion component that is a function of plasma glucoseconcentration, and a dynamic control, i.e., a secretion component that isproportional to the positive values of the glucose concentration derivative. Thestatic component is governed by parameters � and �, whereas the derivativecomponent is governed by parameter kd. Parameter � represents the dynamicdelay between the elevation of glucose concentration and the release of insulinby the pancreas. Parameter � is the proportionality constant between glucoseconcentration and the static component of the secretion response. Parameter kd

is the proportionality constant between the rate of increase of glucose and thederivative component of the secretion response.

Minireview

E378 FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

the dynamic control. The results of the simulation are shown inFig. 10. Figure 10, top, shows the glucose profiles followingeither a square-wave (left) or a meal-like (right) study. Figure10, middle, shows the corresponding model predictions ofinsulin secretion under the assumptions of a normal or adefective �-cell. Below, in Fig. 10, bottom, are the correspond-ing profiles of peripheral insulin concentration. The resultsdeserve some comments. It can be seen that the insulin re-sponse to the square wave of hyperglycemia is similar to theone actually measured during a hyperglycemic clamp (Fig. 10,middle and bottom left). The model predicts a biphasic secre-tory pattern, with a peak of insulin concentration at 3 min, aninterpeak nadir at �10 min, and a subsequent increaseattaining a plateau. The pattern associated with the defective�-cell shows a reduced first phase (which reproduces theresponse of a subject in the initial stage of type 2 diabetes).The insulin-secretory response to the meal-like study (Fig.

10, middle right) shows only a modest sign of biphasicity,i.e., a little bump at 20 min followed by a subsequent surge.The corresponding profile of peripheral insulin concentra-tion increases without any clear sign of biphasicity (Fig. 10,bottom right). In the simulation of the defective �-cell, nosign of biphasicity is present in either insulin secretion orperipheral insulin concentration. Compared with the defec-tive �-cell, the normal �-cell determines a more rapidlyincreasing early phase. In fact, at 20 min, the normal insulinprofile has already achieved 71% of the maximal insulinlevel attained at 60 min, whereas the defective insulinprofile is only at 33% (Fig. 10, bottom right). All in all, itappears that a healthy �-cell, showing a biphasic response toa step increase in glucose concentration, generates an earlyinsulin response to a meal-like study that rapidly elevatesportal and peripheral insulin concentrations. Conversely, adefective �-cell, showing an impaired first-phase insulin

Fig. 10. Results of the model simulationaimed to gain insights into the relationshipbetween the insulin response to a brisk intra-venous glucose challenge and the insulinresponse to a gradual elevation of glucosefollowing food ingestion. The model of Tof-folo et al. (113) was employed to predict�-cell secretion and the corresponding pro-file of peripheral insulin concentration inresponse either to a square wave of hyper-glycemia or to a gradual glucose increasemimicking a meal. The model operated un-der conditions reproducing either a normallyfunctioning or a defective �-cell. Top: glu-cose profiles reproducing either a squarewave of hyperinsulinemia (left) or a meal-like study [right; F, actual average glucoseconcentrations measured during a meal innormal humans (5)]. Middle: correspondingmodel predictions of incremental (abovebasal) insulin secretion under the assump-tions of a normal or a defective �-cell. Bot-tom: corresponding peripheral insulin con-centration profiles. Interpretation of the re-sults is given in the text.

Minireview

E379FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

response to a step increase in glucose concentration, gener-ates a sluggish response to a meal-like study.

There is experimental evidence supporting the conjecturethat the first-phase insulin response to a rapid intravenousglucose challenge and the early insulin response to oral glucoseare emanations of common underlying �-cell mechanisms. Forinstance, the AIRGlc has been shown to be correlated, albeit notstrongly, with the insulinogenic index measured 30 min afterglucose ingestion (the correlation coefficient was 0.61 in Ref.84 and 0.54 in Ref. 54). The correlation between insulinsecretion indexes derived from an IVGTT and a meal hasrecently been confirmed with the minimal-model approach in astudy comparing glucose tolerance in young and elderly sub-jects (4). Also, the first-phase insulin response to the hyper-glycemic clamp and a model-based index of �-cell functionduring the OGTT have been shown to be correlated (11).Another evidence that the insulin responses to intravenous andoral glucose are closely related comes from the observationthat, in patients with IGT or in the early stages of type 2diabetes, the first-phase insulin response to an IVGTT or ahyperglycemic clamp is almost invariably impaired, and thisdefect is mirrored by a more sluggish early insulin responsefollowing glucose ingestion (28, 101) (the role of phasicinsulin release in clinical diabetes is reviewed in Refs. 32, 35,97, and 120).

So far, we have dealt with insulin secretion under physio-logical conditions, focusing on the role of glucose only. Thishas been done because glucose is the most important factoraffecting insulin secretion. However, it is well known thatintravenous glucose elicits a smaller insulin response than oralglucose (42, 53, 83). In this regard, it can be noted that theincremental insulin levels achieved in our simulation of ameal-like study (i.e., a study in which glucose is infusedintravenously in such a way as to mimic the postprandialplasma glucose profile; see Fig. 10) are approximately one-halfof those normally achieved during a real meal (175 vs. 350pmol/l). The increment in insulin response observed after oralcompared with intravenous glucose administration is inter-preted as the presence of other factors besides glucose thatcontribute to stimulate �-cell secretion, such as the neurallyactivated cephalic phase and gastrointestinal hormones (incre-tins). The cephalic phase is a short-lived �-cell secretion that isevoked by the sight, taste, and smell of food (see Refs. 2 and63 and references therein). It precedes food ingestion and ismediated by pancreatic innervation. The cephalic phase causesa rapid (�2 min) but small (�30 pmol/l) increase in plasmainsulin concentration. Albeit modest in absolute terms, thecephalic phase plays a significant physiological role, as aninverse relationship between the cephalic phase and the initialincrease in plasma glucose concentration has been found (17).Incretins are gastrointestinal hormones that have a potentiatingeffect on �-cell secretion after ingestion of a carbohydrate meal(15, 38, 117). It has been suggested that the incretin effect isresponsible for �50% of the insulin response to oral glucose.It follows that incretin deficiency may contribute to the devel-opment of postprandial hyperglycemia and that administrationof exogenous incretin hormones may have therapeutic appli-cations in the treatment of type 2 diabetes. The presence offactors that contribute only to the insulin response to ingestedglucose, but not to the insulin response to intravenous glucose,may explain why the correlation between the AIRGlc and

intravenous glucose and the insulinogenic index measured 30min after glucose ingestion is not strong.

EFFECTS OF EARLY INSULIN RESPONSE ON PRANDIALGLUCOSE METABOLISM

An increasing body of evidence indicates that the earlyinsulin response following glucose ingestion plays a criticalrole in the maintenance of postprandial glucose homeostasis.The early surge in insulin concentration is capable of limitingthe initial glucose excursion mainly through the prompt inhi-bition of endogenous glucose production, with the insulin-mediated curtailment of glucagon secretion being particularlyrelevant (75, 76, 102). This confirms, under physiologicalconditions, the findings of the studies that assessed the meta-bolic relevance of the first-phase insulin response to intrave-nous glucose. In the following discussion, we will provide abrief synopsis of the studies that assessed the benefits associ-ated with the early phase of insulin release as well as theproblems that are associated with its impairment.

We previously mentioned the study by Calles-Escandon andRobbins (18) in relation to the physiological relevance of thefirst-phase insulin response to intravenous glucose. In thatstudy, Calles-Escandon and Robbins also studied the effects ofa selective suppression of the early insulin response (0–30min) to oral glucose by using a brief intravenous infusion ofsomatostatin. They found that the loss of the early phase duringthe OGTT was associated with reduced glucose tolerance. Infact, a transient hyperglycemia that stimulated a sustained lateinsulin response between 60 and 120 min was observed.

Further experimental evidence that the early insulin responseis associated with glucose tolerance comes from a study con-ducted on IGT patients by Mitrakou et al. (75). That studyrevealed an inverse correlation between the plasma insulinlevel achieved 30 min after an oral glucose load and the plasmaglucose concentration measured at 120 min. This confirms thata defective early insulin response may lead to postprandialhyperglycemia, which in turn determines a prolonged stimula-tion of the �-cell leading to hyperinsulinemia.

Basu et al. (3) carried out a study to understand the role ofthe pattern of insulin delivery on the ability to dispose aglucose load. They studied nondiabetic, obese, and diabeticindividuals during a meal-like study in which somatostatin wasused to inhibit endogenous insulin secretion while glucose wasinfused in such a way as to mimic the rate of entry of glucosefrom the gut following the ingestion of 50 g of glucose. Duringthe study, the same amount of exogenous insulin was infusedin such a way as to mimic the postprandial insulin profile ofeither a normal or a diabetic individual, the latter profilefeaturing a sluggish pattern of insulin delivery (correspondingto a loss of early insulin secretion). It was found that the“diabetic” insulin profile was associated with a higher glucosepeak and a prolonged duration of hyperglycemia in all groups(Fig. 11).

Kahn et al. (61) have shown that the relationship betweenthe early insulin response and glucose tolerance is hyperbolicin nature. This result is reminiscent of the hyperbolic relation-ship between the insulin response to intravenous glucose andinsulin sensitivity that was originally suggested by Bergman etal. (8) and then investigated thoroughly by Kahn et al. (60).One important consequence of this nonlinearity is that even

Minireview

E380 FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

small decreases in the early response to glucose ingestion canhave dramatic effects on the postprandial glycemic profile insubjects who are glucose intolerant.

A reduced early insulin response to glucose ingestion isquite a common defect in type 2 diabetes, and this abnormalitycan be a contributing factor in the development and progres-sion of the disease. In fact, in patients with impaired glucosetolerance or in the early stages of type 2 diabetes, the impair-ment of the early insulin response contributes to postprandialhyperglycemia. This maintains the �-cell under a condition ofprolonged stimulation, which eventually leads to late-phasehyperinsulinemia. Chronic postprandial hyperglycemia and hy-perinsulinemia have detrimental effects on the glucose-insulinsystem. Chronic hyperglycemia is capable, through the mech-anism of glucotoxicity (99), of further worsening �-cell secre-tion. Chronic hyperinsulinemia may lead to �-cell exhaustion,may cause downregulation of the insulin receptor thus increas-ing insulin resistance (32), and may produce the well-knownnegative consequences on the vascular endothelium (106, 111).This vicious cycle plays a significant role in the pathogenesisand evolution of type 2 diabetes [as thoroughly examined insome recent surveys (34, 59, 97)]. Thus interventions to restorethe early insulin surge should contribute to improving glucosetolerance in type 2 diabetic patients. In a study by Bruce et al.(16), type 2 diabetic patients received an identical dose ofinsulin in three different regimens at mealtime: insulin admin-istered intravenously over 30 min at the beginning of a meal insuch a way as to mimic a normal early insulin response, withthe same profile but delayed by 30 min, and as a constantinfusion over 60 min. The best result in terms of postprandialglucose tolerance was obtained with the early administration ofinsulin, in keeping with the proposed direct effect of the earlyinsulin response on hepatic glucose production. The beneficial

effect of a short intravenous insulin infusion mimicking theearly insulin response in type 2 diabetic patients has beensubsequently confirmed by Luzio et al. (69).

Because intravenous infusions aimed at achieving an earlyand adequate supply of insulin are clearly impractical, simpleralternatives, applicable to everyday life, have been devised tocorrect a defective early insulin response in diabetic patientswhose disease is inadequately controlled by diet and exercise.One is the subcutaneous administration of fast-acting insulinanalogs; another is the use of pharmacological agents (such asrepaglinide, nateglinide, and similar compounds) that exert apreferential stimulation of the early insulin secretion in thosepatients still having a residual �-cell function. Such therapeuticapproaches have been shown to be effective in reducing post-prandial glucose and insulin excursions in type 2 diabeticpatients, as illustrated in recent review papers by Del Prato andcolleagues (34, 36).

FROM BIPHASICITY TO PULSATILITY

In the previous sections, we argued that the key feature ofinsulin secretion under physiological conditions is not thebiphasic shape but rather the ability to rapidly elevate insulinlevels in the bloodstream. This interpretation is furtherstrengthened by recently obtained insights into the oscillatorynature of insulin secretion following physiological stimuli.When the �-cell response during a meal is being assessed,insulin and C-peptide are usually sampled in the peripheralcirculation every 5–10 min. This is enough to detect thegeneral trend of �-cell secretion and reveal its main features.However, a far more complex and fascinating picture emergeswhen blood samples are collected in the portal vein everyminute, when ultrasensitive insulin assays are adopted, and

Fig. 11. Insulin (left) and glucose (right)concentration profiles measured in the paperby Basu et al. (3). These investigators studiednormal, obese, and type 2 diabetic (NIDDM)subjects during a meal-like study in whichendogenous insulin secretion was inhibitedwith somatostatin and glucose was infused soas to mimic the rate of glucose entry from thegut during a meal. The same amount of ex-ogenous insulin was infused mimicking ei-ther the normal postprandial insulin profile orthe one typically found in type 2 diabeticpatients, in which the early insulin response isimpaired. The latter insulin profile was asso-ciated in all groups with a more elevatedglucose peak and a more prolonged hypergly-cemia. Data are replotted from Basu et al. (3)with the permission of the Journal of ClinicalInvestigation.

Minireview

E381FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

when deconvolution is used to reconstruct insulin secretion, asdone by Pørksen and coworkers in a series of publications [acomprehensive summary of which is provided by the samegroup of researchers in two recent review papers (89, 91)].These investigators have shown that insulin secretion during ameal has an oscillatory nature and consists of a sequence ofpulses whose amplitude and frequency are modulated by glu-cose (the frequency of secretory bursts is �6 min per pulse)(90). Also, the incretin effect appears to be mediated via thepulsatile mode of secretion (92). Thus, even though C-peptideand insulin concentrations sampled in peripheral blood give theimpression that the early �-cell response to a meal increasesgradually, the actual secretion increases with almost an on/offpattern (Fig. 12). This evidence adds further physiologicalrelevance to the studies that investigated the hepatic effects offirst-phase insulin response. In fact, the first-phase insulinresponse to intravenous glucose can be interpreted as themagnification of one of the insulin bursts detected in the portalcirculation in response to a physiological stimulus.

Even though the early secretory pattern during a meal is notas smooth as that predicted by the �-cell model simulation(Fig. 10), the crucial point nevertheless remains the ability ofthe �-cell to generate an insulin response that rapidly exposesthe liver to high insulin levels. The fact that such high insulinlevels have an up-and-down pattern is a bonus that renders

insulin action on the liver more effective. In fact, many studieshave shown that hepatic insulin action is improved wheninsulin is presented to the liver in an oscillatory fashion (asreviewed in Refs. 73, 82, and 91). Interestingly, type 2 diabeticpatients show profound defects in all aspects of pulsatileinsulin release. Alterations of insulin pulsatility in the postab-sorptive state have been identified as a very early sign of �-celldysfunction in IGT and type 2 diabetic patients (74). Afterglucose ingestion, pulses occur less regularly and have signif-icantly lower amplitude in type 2 diabetic patients (91). Giventhe considerable importance of the pulsatile manner of insulinsecretion in optimizing insulin effects on target tissues, resto-ration of the physiological pulsatility of early insulin secretionmight be a valuable therapeutic approach in the management oftype 2 diabetes and influence the future tailoring of antidiabeticdrugs (100). In this respect, it is encouraging that recent dataobtained in healthy subjects have shown that the insulinotropicagent repaglinide, which is used to stimulate early insulinsecretion in type 2 diabetic patients, amplifies the insulin burstmass while preserving insulin pulsatility (58).

SUMMARY

Insulin secretion is not intrinsically biphasic. Insulin secre-tion can occur in a biphasic manner depending on the type andmagnitude of the glucose stimulus. The biphasic shape is afeature that insulin secretion exhibits when the �-cell is ex-posed to a rapidly changing glucose concentration. Underphysiological conditions, when glucose increases gradually,the first and second phases are not clearly distinguishable.

The same dynamic features of the �-cell machinery thatdetermine a pronounced first-phase insulin secretion in re-sponse to a rapid intravenous glucose administration alsoensure a prompt release of insulin into the bloodstream inresponse to a meal. Conversely, the same defect that deter-mines a marked reduction in first-phase insulin secretion in-duces a sluggish increase of insulin in response to a meal.Therefore, even though the IVGTT or the hyperglycemicclamp determine conditions that do not exist in nature, they arevaluable tools that allow investigators to pinpoint defects thatare likely to have an impact on the secretory response duringnormal physiological situations.

Both animal and human studies indicate that the first-phaseinsulin response to intravenous glucose has beneficial effectson the regulation of glucose metabolism. In particular, the firstphase has a profound and long-term inhibitory effect on hepaticglucose production. Likewise, the early insulin response toingested glucose is an important determinant of prandial glu-cose tolerance. The impairment of early insulin release com-monly found in diabetic patients determines a prolonged du-ration of postprandial hyperglycemia and, in turn, a compen-satory late-phase hyperinsulinemia. Therapeutic interventionscapable of restoring the early insulin release ameliorate glucosetolerance in such patients.

Recent studies using frequent portal venous sampling anddeconvolution indicate that insulin secretion following glucoseingestion increases in an oscillatory fashion. Thus insulinsecretion in real life is pulsatile rather than biphasic. Althoughsuch insulin pulsatility is considerably dampened at the sys-temic level, it is likely to exert a priming effect on the liver.

Fig. 12. Pulsatile insulin secretion before and during a meal assessed byintraportal sampling. Portal vein insulin concentrations in the basal state (left)and after meal ingestion (right). The insulin profiles to which the liver isexposed in response to a meal are dynamic and resemble an on/off pattern.Reproduced from Pørksen et al. (90) with the permission of the AmericanJournal of Physiology.

Minireview

E382 FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

This further underlines the notion that insulin shape and func-tion are closely related.

ACKNOWLEDGMENTS

We are grateful to Dr. Elena Breda for help in implementing the �-cellmodel. We thank Prof. Gianna Toffolo and Dr. Marco Campioni for providingthe �-cell model parameters estimated in normal subjects during a meal andduring a meal-like study. We express our gratitude to Prof. Claudio Cobelli formany valuable comments on the manuscript. We also thank Mauryn Martinezfor help with the English revision of the manuscript.

GRANTS

We acknowledge the generous research support provided by Novo-Nordiskin the study of the pathophysiology of insulin secretion.

REFERENCES

1. Ader M and Bergman RN. Peripheral effects of insulin dominatesuppression of fasting hepatic glucose production. Am J Physiol Endo-crinol Metab 258: E1020–E1032, 1990.

2. Ahren B. Autonomic regulation of islet hormone secretion—implica-tions for health and disease. Diabetologia 43: 393–410, 2000.

3. Basu A, Alzaid A, Dinneen S, Caumo A, Cobelli C, and Rizza RA.Effects of a change in the pattern of insulin delivery on carbohydratetolerance in diabetic and nondiabetic humans in the presence of differingdegrees of insulin resistance. J Clin Invest 97: 2351–2361, 1996.

4. Basu R, Breda E, Oberg AL, Powell CC, Dalla Man C, Basu A,Vittone JL, Klee GG, Arora P, Jensen MD, Toffolo G, Cobelli C, andRizza RA. Mechanisms of the age-associated deterioration in glucosetolerance: contribution of alterations in insulin secretion, action, andclearance. Diabetes 52: 1738–1748, 2003.

5. Basu R, Di Camillo B, Toffolo G, Basu A, Shah P, Vella A, Rizza R,and Cobelli C. Use of a novel triple-tracer approach to assess postpran-dial glucose metabolism. Am J Physiol Endocrinol Metab 284: E55–E69,2003.

6. Bergman RN. New concepts in extracellular signaling for insulin action:the single gateway hypothesis. Recent Prog Horm Res 52: 359–387,1997.

7. Bergman RN, Ider YZ, Bowden CR, and Cobelli C. Quantitativeestimation of insulin sensitivity. Am J Physiol Endocrinol Metab Gas-trointest Physiol 236: E667–E677, 1979.

8. Bergman RN, Phillips NLS, and Cobelli C. Physiologic evaluation offactors controlling glucose tolerance in man. Measurement of insulinsensitivity and beta-cell sensitivity from the response to intravenousglucose. J Clin Invest 68: 1456–1467, 1981.

9. Bergman RN and Urquhart J. The pilot gland approach to the study ofinsulin secretory dynamics. Recent Prog Horm Res 27: 583–605, 1971.

10. Bergman RN, Yang YJ, Hope ID, and Ader M. The role of thetranscapillary insulin transport in the efficiency of insulin action: studieswith glucose clamps and the minimal model. Horm Metab Res Suppl 24:49–56, 1990.

11. Bonadonna RC, Stumvoll M, Fritsche A, Muggeo M, Haring H,Bonora E, and Van Haeften TW. Altered homeostatic adaptation offirst- and second-phase beta-cell secretion in the offspring of patientswith type 2 diabetes: studies with a minimal model to assess beta-cellfunction. Diabetes 52: 470–480, 2003.

12. Bratusch-Marrain PR and Waldhausl WK. Impact of the “early peak”of insulin secretion on glucose utilization in man. Horm Metab Res 17:163–164, 1985.

13. Breda E, Cavaghan MK, Toffolo G, Polonsky KS, and Cobelli C.Oral glucose tolerance test minimal model indexes of beta-cell functionand insulin sensitivity. Diabetes 50: 150–158, 2001.

14. Breda E and Cobelli C. Insulin secretion rate during glucose stimuli:alternative analyses of C-peptide data. Ann Biomed Eng 29: 692–700,2001.

15. Brown JC and Otte SC. Gastrointestinal hormones and the control ofinsulin secretion. Diabetes 27: 782–787, 1978.

16. Bruce DG, Chisholm DJ, Storlien LH, and Kraegen EW. Physiolog-ical importance of deficiency in early prandial insulin secretion innon-insulin-dependent diabetes. Diabetes 37: 736–744, 1988.

17. Bruce DG, Storlien LH, Furler SM, and Chisholm DJ. Cephalic phasemetabolic responses in normal weight adults. Metabolism 36: 721–725,1987.

18. Calles-Escandon J and Robbins DC. Loss of early phase of insulinrelease in humans impairs glucose tolerance and blunts thermic effect ofglucose. Diabetes 36: 1167–1172, 1987.

19. Caumo A, Vicini P, Zachwieja JJ, Avogaro A, Yarasheski K, BierDM, and Cobelli C. Undermodeling affects minimal model indexes:insights from a two-compartment model. Am J Physiol Endocrinol Metab276: E1171–E1193, 1999.

20. Cerasi E. Potentiation of insulin release by glucose in man. I. Quanti-tative analysis of the enhancement of glucose-induced insulin secretionby pretreatment with glucose in normal subjects. Acta Endocrinol 79:483–501, 1975.

21. Cerasi E. Differential actions of glucose on insulin release: reevaluationof a mathematical model. In: Carbohydrate Metabolism, edited byCobelli C and Bergman RN. New York: Wiley, 1981, p. 3–24.

22. Cerasi E, Fick G, and Rudemo M. A mathematical model for theglucose induced insulin release in man. Eur J Clin Invest 4: 267–278,1974.

23. Cerasi E and Luft R. Plasma-insulin response to sustained hyperglyce-mia induced by glucose infusion in human studies. Lancet 41: 1359–1361, 1963.

24. Cerasi E and Luft R. The plasma insulin response to glucose infusionin healthy subjects and in diabetes mellitus. Acta Endocrinol 55: 278–304, 1967.

25. Chen M and Porte D Jr. The effect of rate and dose of glucose infusionon the acute insulin response in man. J Clin Endocrinol Metab 42:1168–1175, 1976.

26. Cherrington AD, Edgerton D, and Sindelar DK. The direct andindirect effects of insulin on hepatic glucose production in vivo. Diabe-tologia 41: 987–996, 1998.

27. Cherrington AD, Sindelar D, Edgerton D, Steiner K, and McGuin-ness OP. Physiological consequences of phasic insulin release in thenormal animal. Diabetes 51: S103–S108, 2002.

28. Colwell JA and Lein A. Diminished insulin response to hyperglycemiain prediabetes and diabetes. Diabetes 16: 560–565, 1967.

29. Cretti A, Lehtovirta M, Bonora E, Brunato B, Zenti MG, Tosi F,Caputo M, Caruso B, Groop LC, Muggeo M, and Bonadonna RC.Assessment of beta-cell function during the oral glucose tolerance test bya minimal model of insulin secretion. Eur J Clin Invest 31: 405–416,2001.

30. Curry DL, Bennett LL, and Grodsky GM. Dynamics of insulinsecretion by the perfused rat pancreas. Endocrinology 83: 572–584,1968.

31. Daniel S, Noda M, Straub SG, and Sharp GW. Identification of thedocked granule pool responsible for the first phase of glucose-stimulatedinsulin secretion. Diabetes 48: 1686–1690, 1999.

32. DeFronzo RA. Lilly lecture 1987. The triumvirate: beta-cell, muscle,liver. A collusion responsible for NIDDM. Diabetes 37: 667–687, 1988.

33. DeFronzo RA, Tobin JD, and Andres R. Glucose clamp technique: amethod for quantifying insulin secretion and resistance. Am J PhysiolEndocrinol Metab Gastrointest Physiol 237: E214–E223, 1979.

34. Del Prato S. Loss of early insulin secretion leads to postprandialhyperglycaemia. Diabetologia 46, Suppl 1: M2–M8, 2003.

35. Del Prato S, Marchetti P, and Bonadonna RC. Phasic insulin releaseand metabolic regulation in type 2 diabetes. Diabetes 51, Suppl 1:109–116, 2002.

36. Del Prato S and Tiengo A. The importance of first-phase insulinsecretion: implications for the therapy of type 2 diabetes mellitus.Diabetes Metab Res Rev 17: 164–174, 2001.

37. Eaton RP, Allen RC, Schade DS, Erickson KM, and Standefer J.Prehepatic insulin production in man: kinetic analysis using peripheralconnecting peptide behavior. J Clin Endocrinol Metab 51: 520–528,1980.

38. Ebert R and Creutzfeldt W. Gastrointestinal peptides and insulinsecretion. Diabetes Metab Rev 3: 1–26, 1987.

39. Efendic S, Cerasi E, and Luft R. Quantitative study on the potentiatingeffect of arginine on glucose-induced insulin response in healthy, predi-abetic, and diabetic subjects. Diabetes 23: 161–171, 1974.

40. Elahi D. In praise of the hyperglycemic clamp. A method for assessmentof beta-cell sensitivity and insulin resistance. Diabetes Care 19: 278–286, 1996.

41. Elahi D, Meneilly GS, Minaker KL, Andersen DK, and Rowe JW.Escape of hepatic glucose production during hyperglycemic clamp. Am JPhysiol Endocrinol Metab 257: E704–E711, 1989.

Minireview

E383FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

42. Elrick H, Stimmler L, Hlad CJJ, and Arai Y. Plasma insulin responseto oral and intravenous glucose administration. J Clin Endocrinol Metab24: 1076–1082, 1964.

43. Ferrannini E. Insulin resistance versus insulin deficiency in non-insulin-dependent diabetes mellitus: problems and prospects. Endocr Rev 19:477–490, 1998.

44. Ferrannini E and Cobelli C. The kinetics of insulin in man. I. Generalaspects. Diabetes Metab Rev 3: 335–363, 1987.

45. Ferrannini E and Cobelli C. The kinetics of insulin in man. II. Role ofthe liver. Diabetes Metab Rev 3: 365–397, 1987.

46. Getty L, Hamilton-Wessler M, Ader M, Dea MK, and Bergman RN.Biphasic insulin secretion during intravenous glucose tolerance testpromotes optimal interstitial insulin profile. Diabetes 47: 1941–1947,1998.

47. Giacca A, Fisher SJ, Shi ZQ, Gupta R, Lickley HL, and Vranic M.Importance of peripheral insulin levels for insulin-induced suppression ofglucose production in depancreatized dogs. J Clin Invest 90: 1769–1777,1992.

48. Grodsky GM. A threshold distribution hypothesis for packet storage ofinsulin and its mathematical modeling. J Clin Invest 51: 2047–2059,1972.

49. Grodsky GM. A new phase of insulin secretion. How will it contributeto our understanding of beta-cell function? Diabetes 38: 673–678, 1989.

50. Grodsky GM. Kinetics of insulin secretion. Underlying metabolicevents. In: Diabetes Mellitus: A Fundamental and Clinical Text, editedby LeRoith D and Taylor SI. Philadelphia, PA: Lippincott, 2000, p. 2–11.

51. Grodsky GM, Bennett LL, Smith DF, and Schmid FG. Effect of pulseadministration of glucose or glucagon on insulin secretion in vitro.Metabolism 16: 222–233, 1967.

52. Grodsky GM, Curry D, Landahl H, and Bennett L. Further studies onthe dynamic aspects of insulin release in vitro with evidence for atwo-compartmental storage system. Acta Diabetol Lat 6: 554–578, 1969.

53. Hampton SM, Morgan LM, Tredger JA, Cramb R, and Marks V.Insulin and C-peptide levels after oral and intravenous glucose. Contri-bution of enteroinsular axis to insulin secretion. Diabetes 35: 612–616,1986.

54. Hanson RL, Pratley RE, Bogardus C, Narayan KM, Roumain JM,Imperatore G, Fagot Campagna A, Pettitt DJ, Bennett PH, andKnowler WC. Evaluation of simple indices of insulin sensitivity andinsulin secretion for use in epidemiologic studies. Am J Epidemiol 151:190–198, 2000.

55. Henquin JC, Boitard C, Efendic S, Ferrannini E, Steiner DF, andCerasi E. Insulin secretion: movement at all levels. Diabetes 51: S1–S2,2002.

56. Hovorka R, Chassin L, Luzio SD, Playle R, and Owens DR. Pancre-atic beta-cell responsiveness during meal tolerance test: model assess-ment in normal subjects and subjects with newly diagnosed noninsulin-dependent diabetes mellitus. J Clin Endocrinol Metab 83: 744–750,1998.

57. Hovorka R and Jones RH. How to measure insulin secretion. DiabetesMetab Rev 10: 91–117, 1994.

58. Juhl CB, Porksen N, Hollingdal M, Sturis J, Pincus S, Veldhuis JD,Dejgaard A, and Schmitz O. Repaglinide acutely amplifies pulsatileinsulin secretion by augmentation of burst mass with no effect on burstfrequency. Diabetes Care 23: 675–681, 2000.

59. Kahn SE. Clinical review 135: the importance of beta-cell failure in thedevelopment and progression of type 2 diabetes. J Clin EndocrinolMetab 86: 4047–4058, 2001.

60. Kahn SE, Prigeon RL, McCulloch DK, Boyko EJ, Bergman RN,Schwartz MW, Neifing JL, Ward WK, Beard JC, and Palmer JP.Quantification of the relationship between insulin sensitivity and beta-cell function in human subjects. Evidence for a hyperbolic function.Diabetes 42: 1663–1672, 1993.

61. Kahn SE, Verchere CB, Andrikopoulos S, Asberry PJ, Leonetti DL,Wahl PW, Boyko EJ, Schwartz RS, Newell Morris L, and FujimotoWY. Reduced amylin release is a characteristic of impaired glucosetolerance and type 2 diabetes in Japanese Americans. Diabetes 47:640–645, 1998.

62. Kjems LL, Christiansen E, Volund A, Bergman RN, and Madsbad S.Validation of methods for measurement of insulin secretion in humans invivo. Diabetes 49: 580–588, 2000.

63. Konturek SJ and Konturek JW. Cephalic phase of pancreatic secre-tion. Appetite 34: 197–205, 2000.

64. Lefebvre PJ, Paolisso G, Scheen AJ, and Henquin JC. Pulsatility ofinsulin and glucagon release: physiological significance and pharmaco-logical implications. Diabetologia 30: 443–452, 1987.

65. Lerner RL and Porte D Jr. Relationship between intravenous glucoseloads, insulin responses and glucose disappearance rate. J Clin Endocri-nol Metab 33: 409–417, 1971.

66. Lewis GF, Zinman B, Groenewoud Y, Vranic M, and Giacca A.Hepatic glucose production is regulated both by direct hepatic andextrahepatic effects of insulin in humans. Diabetes 45: 454–462, 1996.

67. Licko V. Threshold secretory mechanism: a model of derivative elementin biological control. Bull Math Biol 35: 51–58, 1973.

68. Luzi L and DeFronzo RA. Effect of loss of first-phase insulin secretionon hepatic glucose production and tissue glucose disposal in humans.Am J Physiol Endocrinol Metab 257: E241–E246, 1989.

69. Luzio SD, Owens DR, Vora J, Dolben J, and Smith H. Intravenousinsulin simulates early insulin peak and reduces post-prandial hypergly-caemia/hyperinsulinaemia in type 2 (non-insulin-dependent) diabetesmellitus. Diabetes Res Clin Pract 16: 63–67, 1991.

70. Maheux P, Chen YD, Polonsky KS, and Reaven GM. Evidence thatinsulin can directly inhibit hepatic glucose production. Diabetologia 40:1300–1306, 1997.

71. Mari A. Mathematical modeling in glucose metabolism and insulinsecretion. Curr Opin Clin Nutr Metab Care 5: 495–501, 2002.

72. Mari A, Camastra S, Toschi E, Giancaterini A, Gastaldelli A, Min-grone G, and Ferrannini E. A model for glucose control of insulinsecretion during 24 h of free living. Diabetes 50, Suppl 1: 164–168,2001.

73. Matthews DR. Physiological implications of pulsatile hormone secre-tion. Ann NY Acad Sci 618: 28–37, 1991.

74. Matthews DR. Oscillatory insulin secretion: a variable phenotypicmarker. Diabet Med 13, Suppl 6: 53–58, 1996.

75. Mitrakou A, Kelley D, Mokan M, Veneman T, Pangburn T, Reilly J,and Gerich J. Role of reduced suppression of glucose production anddiminished early insulin release in impaired glucose tolerance. N EnglJ Med 326: 22–29, 1992.

76. Mitrakou A, Kelley D, Veneman T, Jenssen T, Pangburn T, Reilly J,and Gerich J. Contribution of abnormal muscle and liver glucosemetabolism to postprandial hyperglycemia in NIDDM. Diabetes 39:1381–1390, 1990.

77. Nesher R and Cerasi E. Biphasic insulin release as the expression ofcombined inhibitory and potentiating effects of glucose. Endocrinology121: 1017–1024, 1987.

78. Nesher R and Cerasi E. Modeling phasic insulin release: immediate andtime-dependent effects of glucose. Diabetes 51, Suppl 1: 53–59, 2002.

79. O’Connor MD, Landahl HD, and Grodsky GM. Role of rate ofchange of glucose concentration as a signal for insulin release. Endocri-nology 101: 85–88, 1977.

80. O’Connor MD, Landahl H, and Grodsky GM. Comparison of storage-and signal-limited models of pancreatic insulin secretion. Am J PhysiolRegul Integr Comp Physiol 238: R378–R389, 1980.

81. Pacini G and Mari A. Methods for clinical assessment of insulinsensitivity and beta-cell function. Best Pract Res Clin Endocrinol Metab17: 305–322, 2003.

82. Paolisso G, Scheen AJ, Albert A, and Lefebvre PJ. Effects of pulsatiledelivery of insulin and glucagon in humans. Am J Physiol EndocrinolMetab 257: E686–E696, 1989.

83. Perley MJ and Kipnis DM. Plasma insulin responses to oral andintravenous glucose: studies in normal and diabetic subjects. J Clin Invest46: 1954–1962, 1967.

84. Phillips DI, Clark PM, Hales CN, and Osmond C. Understanding oralglucose tolerance: comparison of glucose or insulin measurements duringthe oral glucose tolerance test with specific measurements of insulinresistance and insulin secretion. Diabet Med 11: 286–292, 1994.

85. Pillonetto G, Sparacino G, and Cobelli C. Reconstructing insulinsecretion rate after a glucose stimulus by an improved stochastic decon-volution method. IEEE Trans Biomed Eng 48: 1352–1354, 2001.

86. Polonsky KS. Lilly Lecture 1994. The beta-cell in diabetes: frommolecular genetics to clinical research. Diabetes 44: 705–717, 1995.

87. Polonsky KS, Given BD, Pugh W, Licinio-Paixao J, Thompson JE,Karrison T, and Rubenstein AH. Calculation of the systemic deliveryrate of insulin in normal man. J Clin Endocrinol Metab 63: 113–118,1986.

Minireview

E384 FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from

88. Polonsky KS, Sturis J, and Van Cauter E. Temporal profiles andclinical significance of pulsatile insulin secretion. Horm Res 49: 178–184, 1998.

89. Pørksen N. The in vivo regulation of pulsatile insulin secretion. Diabe-tologia 45: 3–20, 2002.

90. Pørksen N, Grofte T, Greisen J, Mengel A, Juhl C, Veldhuis JD,Schmitz O, Rossle M, and Vilstrup H. Human insulin release processesmeasured by intraportal sampling. Am J Physiol Endocrinol Metab 282:E695–E702, 2002.

91. Pørksen N, Hollingdal M, Juhl C, Butler P, Veldhuis JD, andSchmitz O. Pulsatile insulin secretion: detection, regulation, and role indiabetes. Diabetes 51: S245–S254, 2002.

92. Pørksen N, Munn S, Steers J, Veldhuis JD, and Butler PC. Effects ofglucose ingestion versus infusion on pulsatile insulin secretion. Theincretin effect is achieved by amplification of insulin secretory burstmass. Diabetes 45: 1317–1323, 1996.

93. Porte D Jr. Banting lecture 1990. Beta-cells in type II diabetes mellitus.Diabetes 40: 166–180, 1991.

94. Porte D Jr and Pupo AA. Insulin responses to glucose: evidence for atwo pool system in man. J Clin Invest 48: 2309–2319, 1969.

95. Poulin RA, Steil GM, Moore DM, Ader M, and Bergman RN.Dynamics of glucose production and uptake are more closely related toinsulin in hindlimb lymph than in thoracic duct lymph. Diabetes 43:180–190, 1994.

96. Prager R, Wallace P, and Olefsky JM. Direct and indirect effects ofinsulin to inhibit hepatic glucose output in obese subjects. Diabetes 36:607–611, 1987.

97. Pratley RE and Weyer C. The role of impaired early insulin secretionin the pathogenesis of type II diabetes mellitus. Diabetologia 44: 929–945, 2001.

98. Rebrin K, Steil GM, Mittelman SD, and Bergman RN. Causal linkagebetween insulin suppression of lipolysis and suppression of liver glucoseoutput in dogs. J Clin Invest 98: 741–749, 1996.

99. Rossetti L, Giaccari A, and DeFronzo RA. Glucose toxicity. DiabetesCare 13: 610–630, 1990.

100. Schmitz O, Brock B, Hollingdal M, Juhl CB, and Pørksen N.High-frequency insulin pulsatility and type 2 diabetes: from physiologyand pathophysiology to clinical pharmacology. Diabetes Metab 28:4S14–4S20, 2002.

101. Seltzer HS, Allen EW, Herron AL, and Brennan MT. Insulin secre-tion in response to glycemic stimulus: relation of delayed initial releaseto carbohydrate intolerance in mild diabetes mellitus. J Clin Invest 46:323–335, 1967.

102. Shah P, Basu A, Basu R, and Rizza R. Impact of lack of suppressionof glucagon on glucose tolerance in humans. Am J Physiol EndocrinolMetab 277: E283–E290, 1999.

103. Shapiro ET, Tillil H, Miller MA, Frank BH, Galloway JA, Ruben-stein AH, and Polonsky KS. Insulin secretion and clearance. Compar-ison after oral and intravenous glucose. Diabetes 36: 1365–1371, 1987.

104. Shapiro ET, Tillil H, Rubenstein AH, and Polonsky KS. Peripheralinsulin parallels changes in insulin secretion more closely than C-peptideafter bolus intravenous glucose administration. J Clin Endocrinol Metab67: 1094–1099, 1988.

105. Sherwin RS, Kramer KJ, Tobin JD, Insel PA, Liljenquist JE, Ber-man M, and Andres R. A model of the kinetics of insulin in man. J ClinInvest 53: 1481–1492, 1974.

106. Simonson DC. Hyperinsulinemia and its sequelae. Horm Metab ResSuppl 22: 17–25, 1990.

107. Sorenson RL, Lindell DV, and Elde RP. Glucose stimulation ofsomatostatin and insulin release from the isolated, perfused rat pancreas.Diabetes 29: 747–751, 1980.

108. Sparacino G and Cobelli C. A stochastic deconvolution method toreconstruct insulin secretion rate after a glucose stimulus. IEEE TransBiomed Eng 43: 512–529, 1996.

109. Steiner KE, Mouton SM, Bowles CR, Williams PE, and CherringtonAD. The relative importance of first- and second-phase insulin secretionin countering the action of glucagon on glucose turnover in the consciousdog. Diabetes 31: 964–972, 1982.

110. Steiner KE, Mouton SM, Williams PE, Lacy WW, and CherringtonAD. Relative importance of first- and second-phase insulin secretion inglucose homeostasis in conscious dog. II. Effects on gluconeogenesis.Diabetes 35: 776–784, 1986.

111. Sundell J and Knuuti J. Insulin and myocardial blood flow. CardiovascRes 57: 312–319, 2003.

112. Toffolo G, Bergman RN, Finegood DT, Bowden CR, and Cobelli C.Quantitative estimation of beta cell sensitivity to glucose in the intactorganism: a minimal model of insulin kinetics in the dog. Diabetes 29:979–990, 1980.

113. Toffolo G, Breda E, Cavaghan MK, Ehrmann DA, Polonsky KS, andCobelli C. Quantitative indexes of beta-cell function during gradedup&down glucose infusion from C-peptide minimal models. Am JPhysiol Endocrinol Metab 280: E2–E10, 2001.

114. Toffolo G and Cobelli C. Insulin modelling. In: Modeling Methodologyfor Physiology and Medicine, edited by Carson E and Cobelli C. London:Academic, 2001, p. 305–335.

115. Toffolo G, De Grandi F, and Cobelli C. Estimation of beta-cellsensitivity from intravenous glucose tolerance test C-peptide data.Knowledge of the kinetics avoids errors in modeling the secretion.Diabetes 44: 845–854, 1995.

116. Unger RH. Glucagon physiology and pathophysiology in the light ofnew advances. Diabetologia 28: 574–578, 1985.

117. Unger RH and Eisentraut AM. Entero-insular axis. Arch Intern Med123: 261–266, 1969.

118. Van Cauter E, Mestrez F, Sturis J, and Polonsky KS. Estimation ofinsulin secretion rates from C-peptide levels. Comparison of individualand standard kinetic parameters for C-peptide clearance. Diabetes 41:368–377, 1992.

119. Volund A, Polonsky KS, and Bergman RN. Calculated pattern ofintraportal insulin appearance without independent assessment of C-peptide kinetics. Diabetes 36: 1195–1202, 1987.

120. Ward WK, Beard JC, and Porte D Jr. Clinical aspects of islet B-cellfunction in non-insulin-dependent diabetes mellitus. Diabetes Metab Rev2: 297–313, 1986.

121. Watanabe RM, Steil GM, and Bergman RN. Critical evaluation of thecombined model approach for estimation of prehepatic insulin secretion.Am J Physiol Endocrinol Metab 274: E172–E183, 1998.

122. Watanabe RM, Volund A, Roy S, and Bergman RN. Prehepaticbeta-cell secretion during the intravenous glucose tolerance test in hu-mans: application of a combined model of insulin and C-peptide kinetics.J Clin Endocrinol Metab 69: 790–797, 1989.

123. Yang YJ, Hope ID, Ader M, and Bergman RN. Insulin transport acrosscapillaries is rate limiting for insulin action in dogs. J Clin Invest 84:1620–1628, 1989.

Minireview

E385FIRST-PHASE INSULIN SECRETION

AJP-Endocrinol Metab • VOL 287 • SEPTEMBER 2004 • www.ajpendo.org

on April 23, 2012

ajpendo.physiology.orgD

ownloaded from