Type 2 Diabetes: The PathologicBasis of Reversible b-CellDysfunctionDiabetes Care 2016;39:2080–2088 | DOI: 10.2337/dc16-0619

The reversible nature of early type 2 diabetes has been demonstrated in in vivohuman studies. Recent in vivo and in vitro studies ofb-cell biology have establishedthat theb-cell losesdifferentiated characteristics, including glucose-mediated insulinsecretion, undermetabolic stress. Critically, theb-cell dedifferentiation produced bylong-term excess nutrient supply is reversible. Weight loss in humans permits res-toration of first-phase insulin secretion associated with the return to normal of theelevated intrapancreatic triglyceride content. However, in type 2 diabetes of dura-tion greater than 10 years, the cellular changes appear to pass a point of no return.This review summarizes the evidence that early type 2 diabetes can be regarded as areversible b-cell response to chronic positive calorie balance.

Until recently, the pathophysiology of type 2 diabetes was believed to be charac-terized by progressive, irreversible loss of pancreatic insulin secretion (1) mediatedby apoptosis of pancreatic b-cells (2). Now that weight loss has been shown to bringabout restoration of b-cell function and reversal of diabetes, a fundamental reap-praisal of the mechanisms underlying b-cell dysfunction is required. The recentin vivo and in vitro studies will be reviewed, together with the evidence that theunderlying, potentially reversible, b-cell failure is related to dedifferentiation ratherthan to b-cell death.

ETIOLOGICAL DRIVERS OF TYPE 2 DIABETES

Population data demonstratemajor increases or decreases in the incidence of type 2diabetes secondary to food excess or scarcity. This was documented in the U.K.during the First and SecondWorld Wars and in Cuba in 1990–96, with an associatedsharp fall in the incidence and prevalence of type 2 diabetes (3,4). Perhaps theclearest evidence for the effect of positive energy balance in those with a predis-posing genotype is provided by the Pima Indians, who had neither excess obesity norexcess diabetes when they lived as subsistence farmers (5,6). In 1940, diabetesprevalence was similar to that of the general U.S. population (7). Thereafter, withcessation of an agricultural lifestyle, together with food oversupply, a dramaticincrease occurred in the rates of obesity, and the prevalence of type 2 diabetes inadult Pima Indians rose to 38% (8). There was only a modest rise in the prevalence inethnically identical Pima Indians living in Mexico under nutritional conditions thatlimit adult weight gain (8).Type 2 diabetes is commonly said to be a consequence of obesity. However, in the

1970s, when the averageweight of the U.K. populationwas considerably less than atpresent, the Whitehall study showed only a small association between obesity andtype 2 diabetes (9). At that time it was considered that obesity had no major effecton the development of common type 2 diabetes (9–11). Although this may seem

Regenerative Medicine for Diabetes Group andMagnetic Resonance Centre, Institute of CellularMedicine, Newcastle University, Newcastle uponTyne, U.K.

Corresponding author: Roy Taylor, roy.taylor@ncl.ac.uk.

Received 21March 2016 and accepted 23 August2016.

© 2016 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered.More infor-mation isavailableathttp://www.diabetesjournals.org/content/license.

Michael G. White, James A.M. Shaw, and

Roy Taylor

2080 Diabetes Care Volume 39, November 2016

REV

IEW

surprising from a present day perspective,the risk of type 2 diabetes rises moststeeply at very high BMIs, and only 7%of the population had a BMI.30 kg/m2

in 1980 (12). The effect of high BMI onthe development of diabetes was simplynot detectable. Although this associationwith obesity is now apparent, it has notgenerally been recognized that diabetesrisk rises steadily throughout the popu-lation weight distribution. The Nurses’Health Study showed a fourfold increasein type 2 diabetes prevalence in womenwith BMI of 23–25 kg/m2 comparedwith those with a BMI of,22 kg/m2 (13).In the UK Prospective Diabetes Study,which recruited between 1977 and 1991,36% of newly diagnosed individualshad a BMI of,25 kg/m2 (14). Conversely,72% of people with a BMI.40 kg/m2 donot have diabetes (15).The individual susceptibility to type 2

diabetes that clusters in families mustbe considered. The condition does notoccur unless b-cell function is no longersufficient to overcome insulin resistance(16). Genome-wide association studiesindicate that the vast majority of thegenes associated with type 2 diabetesare likely to have a b-cell–specific rolerelating to impaired ability to cope withmetabolic stress (17). Although themolec-ular mechanisms remain to be defined, arange of susceptibility to b-cell damagelikely exists, with differences betweenindividuals.

THE PATHOLOGIC BASIS OFDISEASE PROGRESSION

At the diagnosis of type 2 diabetes,b-cell function is typically reduced to50% of normal by HOMA modeling andto a greater extent on dynamic testing(1,18). Despite the initial effect of dietand oral therapy to lower glucose, ob-servational studies have shown thatdisease progression is associated withinexorably declining b-cell function andprogression to insulin commencement,with relatively minor changes in under-lying insulin resistance. Such observa-tions have been made in the context ofcontinuing weight gain (19).Metabolic studies have enabled eluci-

dation of dynamic changes in b-cell func-tion and insulin resistance over time. Incontrast, the absence of imaging modali-ties with sufficient resolution to imagethe b-cell in situ has prevented determi-nation of whether these changes are

caused by dysfunction or true loss ofb-cell mass at a cellular level (20). More-over, satisfactory circulating markers ofb-cell death or proliferation are currentlylacking (21).

Postmortem pancreatic pathologystudies, based on presence of stainingfor cells that contain insulin, have sug-gested that b-cell mass is significantlyreduced in type 2 diabetes in comparisonwith age-, sex-, and BMI-matched controlindividualswithout diabetes (2). Althoughit is accepted that increased apoptosisplays a role in decreased b-cell massover time, pancreatic pathology findingsin a large cohort of European subjects in-dicate that apoptosis alone is insufficientto explain the profound islet dysfunctionin established type 2 diabetes (22). Otherfactors must contribute to the describeddecrease in cells that stain positive forinsulin in the pancreatic islets.

GLUCOLIPOTOXICITY

In diabetes-prone rodent models oftype 2 diabetes, the b-cell fails duringoverfeeding in those with the geneticpredisposition, and this genetic suscep-tibility to lipid availability is reflected instudies of isolated islets (23). Whenfatty acid concentrations are elevatedin vitro, lipid synthesis and storagewithin the b-cell is favored, and chronicexposure of the b-cell to fatty acid ex-cess directly impairs glucose-stimulatedinsulin secretion (Fig. 1) (24–26). Pro-longed exposure to elevated levels offatty acids in vitro directly results inb-cell stress and dysfunction (27).

Exposure of the rat insulinoma b-cellline to oleic acid brings about storage inintracytoplasmic vacuoles, whereas thesaturated fatty acid palmitate inducesexpansion of the endoplasmic reticulum(Fig. 1A and B) (27). This is known to beassociated with markers of endoplasmicreticulum stress, which are typically el-evated in human b-cells from individu-als with type 2 diabetes (28–30). Themore physiologic exposure to mixed sat-urated and unsaturated fatty acids de-creases insulin secretion, and subsequentremoval of fatty acid from the mediumallows the return of insulin secretionover 24 h (Fig. 1C and D) (27). Humanislets also take up fatty acids avidly,and incubation in 0.33mmol/L palmitateleads to a very large increase in islet tri-glyceride content associated with mark-edly impaired function (Fig. 1E and F)

(25). Once hyperglycemia occurs, theconcomitant elevated glucose is likelyto compound the metabolic insult (31).

Despite the growing body of in vitroand in vivo animal data (31) that supportan important role of glucolipotoxicityin type 2 diabetes pathogenesis andprogression, human studies have provedmore challenging. Hyperglycemia re-versibly impairs insulin secretion in vivo(32,33). The combination of hyperglyce-mia and raised plasma free fatty acidshas an additive effect (34). Prolongedintralipid infusion severely impairsb-cell function in subjects predisposedto develop type 2 diabetes (35). Removalof excess lipid from the pancreas (by de-creased supply in the face of continuingoxidation for energy needs) at the sametime as decreasing plasma VLDL1-triglyc-eride allows the return of normal insulinsecretion in early type 2 diabetes (36,37).A recent study that used optimized pan-creaticMRI showed the removable excessof intrapancreatic triglyceride by weightloss was specific to type 2 diabetes (38).

b-CELL DEDIFFERENTIATION

Loss of the fully differentiated pheno-type is a recently recognized potentialmechanism underlying loss of b-cellfunction in type 2 diabetes (38–43).The metabolic stress of chronic nutrientoversupply can lead to reduced expres-sion or nuclear activity of key b-cell tran-scription factors, including Pdx1, Nkx6.1,and MafA (39–41). This results in loss ofcritical end-differentiated genes, includ-ing insulin itself, in parallel with inductionof “disallowed genes,” including lactatedehydrogenase and hexokinase (44).b-Cell dysfunction ensues from a combi-nation of decreased insulin biosynthesisand lossof physiological nutrient-secretioncoupling. This has been demonstratedin a number of models of glucotoxicity,including partial pancreatectomy, withchanges largely prevented by mainte-nance of normoglycemia (45), and inb-cells in vitro after chronic palmitate ex-posure (46).

Accili and colleagues (47) proposedan important role of the FoxO transcrip-tion factors in type 2 diabetes pathogen-esis. Initially, enhanced FoxO1 nucleartranslocation appears to maintain theactivation state of a subset of b-cell tran-scription factors, including MafA, pre-serving glucose oxidation, suppressingfatty acid oxidation, and thus limiting

care.diabetesjournals.org White, Shaw, and Taylor 2081

mitochondrial stress. Through this mech-anism, FoxO1 is able to orchestrate a com-pensatory response aimed at preservingb-cell function under metabolic stress(Fig. 2). Establishment of chronic hypergly-cemia,however, leads toFoxO1degradationand b-cell decompensation with impairedcapacity for glucose oxidation and increasedability to oxidize palmitate. This has beendescribed as “metabolic inflexibility” with

generationof toxic products, includingper-oxides, and impaired insulin secretion (47).

Lineage-tracing studies in mice withb-cell–specific deletion of FoxO1 ex-posed to metabolic stressors, includingaging and multiple pregnancies, demon-strated that loss ofb-cell mass was causedby dedifferentiation rather than bydeath (41). While maintaining expressionof the endocrinemarker chromogranin A,

expression of insulin was lost in parallelwith a number of key b-cell transcrip-tion factors (PDX1, Nkx6.1, and MafA).Furthermore, lineage-traced dedifferen-tiated cells (also referred to as “empty”cells) expressed a number of genes notnormally associated with adult b-cells, in-cluding mesenchymal markers (vimentin)and pancreatic progenitor markers suchas neurogenin 3 (Fig. 2). Relevance totype 2 diabetes was evidenced by thedemonstration that progressive loss ofFoxO1 was associated with a large num-ber of chromogranin A+/insulin2 cells inrodent models of type 2 diabetes. Giventhe inability to directly assess cellularpathologies in people with diabetes,translation of these studies to an under-standing of human pathophysiology hasbeen challenging. Although b-cells can-not be genetically labeled for in situ line-age-tracing studies in humans, rodentmodels (39,41,48) have enabled identifi-cation of specific transitional phenotypesthat can be stained for in pathological pan-creatic samples. These can help determinethe characteristics and fate of humanb-cells during the course of clinical diabetes.

Supported by detailed quantitativehuman pathology analysis, it is now es-tablished that impaired b-cell functionin type 2 diabetes cannot be accountedfor by increased apoptosis alone (22,49).Recent postmortem studies demon-strate that b-cell dysfunction in type 2diabetes may be associated with degranu-lation and alterations in the b-cell pheno-type. Expression of keyb-cell transcriptionfactors, including PDX1 and MafA, ismarkedly reduced in type 2 diabetes, sug-gesting that dedifferentiation (definedby loss of canonical b-cell markers) maycontribute to impaired b-cell function(50,51). Furthermore, expression of disal-lowed genes, including vimentin, hasbeen detected in b-cells from patientswith type 2 diabetes, providing supportfor the observations in rodents made byTalchai et al. (41). In a recent pancreaspathology study, “empty”b-cellswere as-sessed by coimmunofluorescent stainingfor the endocrine secretory granulemarker, synaptophysin, and all endocrinehormones (insulin, glucagon, somato-statin, and pancreatic polypeptide) (52),as described in db/db and GIRKO mice.Through use of this approach andnormal-ization to b-cell number, the study deter-mined that the number of “empty” ordedifferentiated endocrine cells was

Figure 1—Interaction of fatty acids withb-cell ultrastructure and function. A: Effects of exposureto 0.33 mmol/L oleate upon the ultrastructure of rat insulinoma cells. Triglycerides accumulatein round-shaped droplets in the cytoplasm (arrows). B: Effects of exposure to 0.33 mmol/L palmi-tate upon the ultrastructure of rat insulinoma cells. Triglycerides formed splits in the cytoplasmadjacent to the endoplasamic reticulum (arrows). Ins, insulin granules; N, nucleus. Scale bars =1 mm. Original photomicrographs provided by Drs. Katherine Pinnick and Anne Clark (27). C:Exposure of mouse islets to oleate or palmitate (0.5 mmol/L) for 48 h significantly impairedglucose-stimulated insulin secretion (data from Pinnick et al. [27]). D: Impairment in glucose-stimulated insulin secretion in mouse islets brought about by incubation for 72 h in a fatty acid(FA) mixture of oleate and palmitate (1:1; 0.5 mmol/L) was not present after 48-h exposure,followed by 24 h in fatty acid–free media (nil) (reproduced with permission from Pinnick et al.[27]). *P, 0.05. Effect of exposure of human islets to 0.33 mmol/L palmitate on islet triglyceridecontent (E) and islet glucose-medicated insulin secretion (F). Graphs E and F are reproduced withpermission from Lalloyer et al. (25).

2082 Reversible b-Cell Dysfunction Diabetes Care Volume 39, November 2016

significantly increased in people withtype 2 diabetes compared with controlsubjects without diabetes (31.0% vs.8.7%). Using a similar approach, Butleret al. (53) determined that although al-tered b-cell phenotypes were evident,this did not account for the b-cell deficitin type 2 diabetes. However, expressionof b-cell transcription factors was notassessed within their study, thus limit-ing complete determination of b-cell(de)differentiation status. In contrast,Cinti et al. (52) were able to demonstratethat a number of b-cells displayed al-tered subcellular localization of Nkx6.1and MafA, with expression almost exclu-sively within the cytoplasm. These find-ings are consistent with work by Spijkeret al. (40) and indicate a functional

impairment (caused by cellular localiza-tion) of these key transcription factorsin human diabetes. The extent to whichsuch alterations contribute to b-celldysfunction remains to be determined.However, further work exploring the as-sociation between the stage of dediffer-entiation and functionality is likely to yieldkey mechanistic insights.

Studies exploringb-cell dedifferentiationas a mechanism underlying dysfunctionhave, to date, been unable to elucidatethe specific underlying signaling path-ways. For example, it is not possible todetermine separately the effect of hyper-glycemia and dyslipidemia on b-cell(de)differentiation status in db/db mice orin pathological samples from human do-nors with diabetes. Recent studies have

sought to explore the effects of indi-vidual stresses. For example, nutrient-stimulated insulin secretion wasprevented in a transgenic mouse modelwith an activating KATP channel mutationof b-cells (39). Hyperglycemia in thismodel was associated with b-cell dedif-ferentiation, characterized by the loss ofthe end-differentiated b-cell markersPdx1 andMafA. In an ex vivo human isletlipotoxicity model, incubation with pal-mitate led to b-cell dysfunction associatedwith loss of the end-differentiated phe-notype evidenced by reduced levels ofkey b-cell transcription factors (46).

HYPERGLUCAGONEMIA IN TYPE 2DIABETES

Type 2 diabetes has long been recog-nized to be a dual hormonal disorder,with stimulation of hepatic glucose pro-duction by aberrant postprandial hy-perglucagonemia playing an importantrole in overall hyperglycemia (54,55).Whether this state of relative glucagonoversecretion is driven by increaseda-cell mass or simply increased functionremains unknown. Henquin et al. (56)observed no increases in a-cell mass inpatientswith type 2 diabetes, concludingthat relative hyperglucagonemia waslikely being driven by reduced b-cellmass and subsequent alterations inthe inhibitory actions b-cells exert ona-cells. In contrast, Yoon et al. (57) ob-served that reduced b-cell mass was as-sociated with expansion of the a-cellmass in a different cohort of patientswith type 2 diabetes. Although it wasproposed that increased a-cell massmay be a driver of hyperglucagonemia, arole for reduced b-cell paracrine signalingcannot be excluded in this study.

Given the limitations such cross-sectional human studies pose, definitiveevidence for changes in a-cell mass havelargely been provided by rodent studies,with recent evidence proposing a poten-tial role for b-cell–to–a-cell conversionas a mechanism underlying this phenom-enon (39). Through use of the Fox01 ab-lation model, Talchai et al. (41) were ableto demonstrate that, after metabolicstress, an increase in plasma and pancre-atic glucagon was associated with b-cell–to–a-cell conversion, as determined byglucagon expression in b-derived cells.In support of this, and a role for hypergly-cemia in these plasticity events, Breretonet al. (39) describe b-cell–to–a-cell

Figure 2—Schematic representation of possible stages of b-cell fate changes associated withdiabetes progression. During metabolic stress, FoxO1 translocates to the nucleus to orchestratea compensatory response to help preserve glucose oxidation and suppress fatty acid oxidation.FoxO1 target genes include b-cell transcription factors MafA and NeuroD, which through con-tinued activation of insulin expression help preserve b-cell function under metabolic stress.Continued stress leads to loss and/or cytoplasmic translocation of b-cell transcription factors(Nkx6.1 and MafA) and reduced/lost insulin expression. Analysis of human and rodent dataindicates that cells, at this stage of diabetes, can undergo significant plasticity events thatinvolves the (re)expression of “disallowed genes.” At this stage, cells can be characterized bytwo distinct identities: 1) hormone “empty” cells that may express progenitor (Ngn3 and Oct4)and mesenchymal (vimentin) proteins and 2) transdifferentiated/bihormonal cells that expressother endocrine hormones, including glucagon. Preclinical data indicate that recovery of b-cellfailure at this stage of diabetes can be achieved through redifferentiation of dedifferentiated/transdifferentiated cells, a phenomenon that has been implied clinically also after a calorie-restricted diet. Sustained metabolic stress may result in irreversible b-cell failure throughapoptosis (dedifferentiated cells) and/or established a-cell differentiation; however, this hasnot yet been confirmed in humans. The solid arrows indicate observed rodent and humanphenotypes, and the broken arrows indicate potential cellular pathways that may account forb-cell loss and/or recovery.

care.diabetesjournals.org White, Shaw, and Taylor 2083

conversion in the KATP transgenic mousemodel. Although describing similar fate-switching events, this study observed anumber of bihormonal cells (insulin/glucagon coexpressing), indicating adirect conversion. In contrast, Talchaiet al. (41) observed that the expressionof glucagon in b-derived cells only oc-curred once insulin was lost, with no co-expression evident. These observationsmay possibly be explained by the differ-ences in each of the transgenic modelsused and subsequent variations in the na-ture and timing of b-cell stress. What re-mains consistent, however, is that bothstudies provide evidence for a shift in en-docrine phenotype after the loss of keyb-cell transcription factors.b-Cell dedifferentiation, characterized

by the loss of b-cell–specific markers, in-cluding Pdx1, Nkx6.1, and MafA, mayenable conversion toward other pancre-atic endocrine phenotypes that are lesssusceptible to nutrient-induced mito-chondrial and endoplasmic reticulumstress (58). This raises the possibility thatb-cell identity may be fragile andthat metabolic stress eliminates factorsthat normally repress non–b-cell identi-ties. In support of this, deletion of Pdx1in adult rodents leads to severe hyper-glycemia in tandem with ultrastructuraland physiological a-cell characteristicsin a large fraction of the b-cells (48).These data are in line with b-cell–to–a-cell fate-switching events after loss ofb-cell–specific transcription factors subse-quent to chronic hyperglycemia (39). Col-lectively, these studies indicate thata-cell and potentially other endocrinecell phenotypes may be “default” line-ages and that loss of criticalb-cell factorsthat suppress non–b-cell–related genes,including glucagon, during type 2 diabe-tes pathogenesis may lead to a shift inendocrine phenotype from b-cell toa-cell (39,41,48). In light of recent datafrom Marroqui et al. (58) demonstratingthat a-cells are more resistant to meta-bolic stress, it is possible that the transi-tion from b-cell to a-cell during diabetesmay be a defense mechanism to main-tain cellular mass (58). Because reversalof type 2 diabetes is associatedwith a fallto normal of fasting plasma glucagonlevels at the same time as the return ofnormal b-cell function, the in vivo hu-man studies may be reflecting rediffer-entiation of b-cells. No change in plasmaglucagon is seen in people with normal

glucose tolerance during similar weightloss (59).

In support of earlier work by Whiteet al. (43), a recent study observed aneightfold increased frequency of insulin+

cells coexpressing glucagon in tissue ob-tained from patients with type 2 diabe-tes, indicating such b-cell plasticityoccurs in human diabetes (40). Thestudy determined that half of the bihor-monal (insulin+/glucagon+) cells werenegative for Nkx6.1. Furthermore, cyto-plasmic localization of Nkx6.1 wasobserved, a phenotype not evident inpatients without diabetes. This was con-firmed by the demonstration that a num-ber of cells with cytoplasmic Nkx6.1expression coexpressed glucagon (52),indicating a role for Nkx6.1 in repressingthe a-cell program, as described in vitro(60). Nkx6.1+/glucagon+/insulin2 cellswere also identified, indicating that fac-tors other thanNkx6.1 are critical to thesephenotypic transitions (40).

Given the nature of these cross-sectional studies, interpretation mustbe cautious. For example, it cannotbe ruled out that bihormonal cells reflectb-cell neogenesis even though this ap-pears unlikely. b-Cell neogenesis hasbeen described as originating close toand within ducts (2,61), and associatedbihormonal cells were localized to singleand small clusters of b-cells rather thanwithin established islets (61). Humanstudies using isotopic incorporationinto DNA or lipofuscin accumulation in-dicate no significant b-cell neogenesisoccurs in adults (62,63). Also, Nkx6.1 ismislocalized (cytoplasm) in glucagon+

cells (52). Although these postmortemstudies cannot provide definitive evi-dence, a contributory role for b-cell con-version to a-cell in hyperglucagonemia inhuman diabetes is supported (Fig. 2).

LESSONS ON b-CELL PLASTICITYFROM REVERSING TYPE 2DIABETES

The Counterpoint study has providedthe clearest data on the time course ofrecovery of b-cell function after calorierestriction in type 2 diabetes. In thisstudy, observations were made at 1, 4,and 8 weeks after commencing a verylow-calorie diet (36). After 1 week, therewas no improvement in the first-phaseinsulin response to a stepped insulin se-cretion test with arginine. This is espe-cially significant because fasting plasma

glucose had already normalized as a con-sequence of the very rapid return ofnormal hepatic insulin sensitivity. Hence,the glucotoxicity component had beenremoved and could no longer be ex-erting a major effect in suppressing thefirst-phase insulin response. The effectof raised glucose concentrations in inhib-iting insulin secretion is known to be rap-idly induced and removed (32). Similarly,other suggested mechanisms of stress-induced b-cell dysfunction should berapidly reversible. For instance, the ap-parent mitochondrial dysfunction oftype 2 diabetes is seen only when fastingplasma glucose is.8 mmol/L (64) and israpidly corrected by suppression ofplasma fatty acid levels (65). Reversalof oxidative stress and endoplasmic re-ticulum stress is also associated withrapid restoration of normal b-cell func-tion (27,66).

By 4 weeks into the Counterpointstudy, a first-phase insulin responsecould be seen in the group as a whole,and by 8 weeks, this was well within thenormal range and significantly improvedfrom baseline. The slow normalizationof the first-phase insulin response over8 weeks was mirrored by a slow normal-ization of total intrapancreatic fat (36).During this extended interval, the intra-pancreatic triglyceride concentrationgradually declined to the same level asin control subjects without diabetes(36). The parallel time courses of thefall in excess triglyceride within the pan-creas and the recovery of b-cell functionare suggestive but not conclusive ofcause and effect.

Whether the fall in intrapancreatictriglyceride content could merely reflectthe very considerable weight loss andnot be related to restoration of b-cellfunction had to be considered. A furtherclinical study was conducted to deter-mine whether weight loss also broughtabout a fall in intrapancreatic triglycer-ide content in normoglycemic individu-als or whether it was specific to type 2diabetes. The intrapancreatic triglycer-ide content in a group about to undergoweight loss by gastric bypass surgerywas quantified before and 8 weeks afterthe operation (38). Aweight loss of;13%occurred in those with normal glucosetolerance as well as in those with type 2diabetes. The former showed no changein intrapancreatic triglyceride, whereasthose with type 2 diabetes had higher

2084 Reversible b-Cell Dysfunction Diabetes Care Volume 39, November 2016

levels at baseline, which fell to normal. Atthe same time, the first-phase insulin re-sponse was restored to normal. Thisstudy demonstrates that there is an in-creased pool of triglyceride within thepancreas in people with type 2 diabetesand that substantial weight loss is associ-ated with clearance of this triglycerideexcess. As discussed above, exposure ofb-cells to excess saturated fatty acidcauses avid uptake and decreases the in-sulin secretory response to a change inglucose concentration. It appears possi-ble that the gradual clearance of excesstriglyceride from the pancreas might becausally linked with the recovery of insu-lin secretion. Quantitative comparisonwith measurements on isolated isletsin vitro is not currently feasible.Weight loss of 15 kg in individuals

with type 2 diabetes leavesmany peoplestill in the obese category. It has to beconsidered that fat removed from theliver and the pancreas by short-term hy-pocaloric dieting might gradually be re-placed from the remaining excess insubcutaneous and visceral depots. Ifso, follow-up might be hypothesized toreveal reaccumulation of intrapancre-atic fat with or without a decline inb-cell function. This has been examinedby 6 months of follow-up after acuteweight loss of people with type 2 diabe-tes (37). The group that achieved a fast-ing plasma glucose of ,7 mmol/L afterweight loss demonstrated intrapancre-atic triglyceride content falling to nor-mal levels. Critically, weight remainedstable over 6 months. The first-phase in-sulin response became and remainednormal in this group. There was no re-accumulation of fat in the pancreas orliver even though the mean BMI was30 kg/m2 (37).Return to normal blood glucose con-

trol after weight loss is strongly relatedto duration of type 2 diabetes. Whereas87% of a short-duration group (,4years) achieved nondiabetic fastingplasma glucose levels immediately afteracute weight loss, only 50% of a long-duration group (8–23 years) did so(67). In an audit of outcomes after bari-atric surgery, HbA1c of ,43 mmol/mol(6.1%) was achieved by 62% and 26%,respectively, in those with durationof diabetes of ,4 or .8 years, respec-tively (61), reflecting previous observa-tions (68). The Scandinavian ObesityStudy also confirmed the importance

of duration of diabetes in rates of rever-sal of diabetes at 2 years (69). As theduration of diabetes increases, it ap-pears that a point of no return is passedwith progression to fully differentiatedalternative endocrine lineages (eitherdirectly or after dedifferentiation) and/or irreversible apoptosis (Fig. 2).

Redifferentiation provides an attrac-tive potential underlying mechanism forrecovery of b-cell function after a reduc-tion in islet fat content over the timecourse observed in vivo in humans. It canbe hypothesized that this time courseis too slow for reversal of acute stress-induced dysfunction and too fast forb-cell neogenesis. The dedifferentiatedstate could be considered as providing a“hideaway” until the metabolic insult sub-sides, offering an opportunity for resto-ration of the end-differentiated stateafter alleviation of metabolic stresses.This “therapeutic window” appears lim-ited given the failure to rescue b-cellfunction in subjects with long standing di-abetes (59). It indicates the potential pro-gressive nature of dedifferentiation, witha point of no return in (de)differentiationstatus and/or apoptosis and subsequentirreversibility of loss of function/mass. Anassociation may be postulated betweenthe extent of these abnormalities, thestage of diabetes, and the potential forreversal through b-cell redifferentiation(Fig. 2).

The potential for redifferentiationas a mechanism underlying restorationof b-cell function in type 2 diabetes isillustrated by the restoration of b-cellfunction by intensive insulin therapyin a rodent model of hyperglycemia(42). This restoration in b-cell functionwas associated with b-cell redifferentia-tion characterized by increased levels ofmature b-cell markers, including insulin,Pdx1, and MafA. These observationssupport the proposition that after re-moval of hyperglycemia and change inlipid metabolism achieved by intensiveinsulin therapy, b-cell redifferentiationcan occur and contribute to diabetes re-versal. These studies were extended todemonstrate a restoration in sulfonyl-urea-sensitive insulin secretion, similarto that observed in patients with type 2diabetes after intensive insulin therapy.

It is striking that strategies targetingrestored pancreatic insulin secretion di-rectly have not been shown to be diseasemodifying by restoring functional b-cell

mass, although the A Diabetes OutcomeProgression Trial (ADOPT) study demon-strated greater preservation of b-cellfunctionwith the insulin sensitizer rosigli-tazone compared with sulfonylurea se-cretagogue therapy. Use of exenatideover 3 years caused considerable weightloss and brought about some return ofb-cell function (70). Overall, decreasedb-cell exposure to nutrient excess ap-pears to facilitate reversal of acute dys-function and adaptive dedifferentiation(44). Specific therapies targeted towardreversal of b-cell dedifferentiation areneeded (71–73). The major question ofhow function might be returned after ap-parent end-stage dedifferentiation needsto be addressed.

Identification is underway of candi-date pathways that could be targetedto reverse b-cell dedifferentiation asso-ciated dysfunction (74). A small mole-cule inhibitor of the transforminggrowth factor-b receptor (Alk5) wasshown to be capable of reversing b-celldedifferentiation in islets isolated frommice with extreme diabetes, with uro-cortin 3 used as a marker for matureb-cells. Potential translation to humandisease was indicated by incubation ofisolated human islets with ALK5 inhibi-tor, bringing about increased expressionlevels of a number of mature b-cellmarkers, including insulin, Pdx1, andMafA. However, administration ofALK5 inhibitor to diabetic animals failedto improve glycemic control and causedan overall deterioration in health. Thishighlights the requirement for develop-ment of pathway-specific therapeuticsto restore mature, functional b-cellsif a successful pharmacological ap-proach to controlling type 2 diabetes isto be developed.

A UNIFYING HYPOTHESIS

The 2008 twin cycle hypothesis could po-tentially allow synthesis of the in vivo andin vitro observations (75). The hypothesispostulated that a chronic positive caloriebalance, in the presence of preexistingperipheral insulin resistance and, hence,hyperinsulinemia, would bring aboutsteady accretion of intrahepatic triglycer-ide and initiate linked vicious cycles (Fig.3). Although the mechanisms underlyingthe liver cycle are well established, therecent in vitro work on b-cell dedifferen-tiation provides an attractive potentialexplanation for the operation of the

care.diabetesjournals.org White, Shaw, and Taylor 2085

pancreas cycle (75). The increased plasmaVLDL-triglyceride of type 2 diabetes ispostulated to bring about fat accumula-tion and b-cell stress, such that plasmaglucose remains elevated for longer aftermeals. The consequent effects of over-supply of saturated fatty acids on b-celldedifferentiation could cause graduallydecreasing ability to mount an acute in-sulin response to eating. The liver andpancreas cycles mutually interact and re-inforce each other. At a critical point, theb-cell will fail, resulting in type 2 diabetes.All of these predictions of the 2008 hy-pothesis have now been observed in vivo(36–38).Although most people who develop

type 2 diabetes are overweight or obeseby BMI criteria, the condition is not un-common in thosewithin thenormal range.In the UK Prospective Diabetes Study pop-ulation, 36% had a BMI of,25 kg/m2 at a

time when 64% of the whole U.K. popula-tion had a normal BMI (12). The propor-tion of people with type 2 diabetes whoare of normal BMI has decreased as theprevalence of overweight/obesity has in-creased, caused by the exponential rela-tionship between BMI and the prevalenceof type 2 diabetes. Between 2000 and2008, 11.3% those newly diagnosed withtype 2 diabetes had a BMI of ,25 kg/m2

comparedwith 35% in the entire U.K. pop-ulation (76). Direct observational data sug-gest that a threshold effect operatesdthepersonal fat threshold conceptdwherebyectopic fat accumulationonly occurswhenan individual’s capacity to store fat safelyin the subcutaneous compartment is ex-ceeded (37,77). BMI, which was originallydeveloped as a population metric, maymislead when applied to the individual iftheir adult weight gain is within the nor-mal range. The concept could also explain

the predisposition of some ethnic groups,notably south Asians, to develop type 2diabetes at a relatively low BMI.

The recent observations upon lack ofrestoration of the first-phase insulin re-sponse in long-duration type 2 diabetesafter substantial weight loss (37) cannow be explained by b-cell dedifferenti-ation. The process of losing the special-ized function to produce insulin can bepostulated to reach a point beyondwhich removal of the initial insult of ex-cess fatty acid is not followed by therestoration of nuclear expression ofkey b-cell–specific transcription factorsand recovery of differentiated glucose-responsive insulin secretion (Fig. 2). Thedata discussed in this review allow a sim-plification of understanding of type 2 di-abetes, and this will allow reassessmentof the many associated processes knownto be abnormal.

Type 2 diabetes may now be seen as apotentially reversible state associatedwith longstanding nutrient overload insusceptible individuals. The b-cell dys-function and loss of end-differentiatedb-cell phenotype can be restored by sub-stantial weight loss. After;10 years, theonward march of b-cell dedifferentiationappears likely to precipitate irreversibleloss of insulin secretion unless a substan-tial decrease in body weight is achieved.

Acknowledgments.Theauthors are grateful toDrs. Anne Clark and Katherine Pinnick of OxfordUniversity, Oxford, U.K., for permission to pub-lish the original electron micrographs in Fig. 1Aand B, and to Prof. Bart Staels, of The InstitutePasteur de Lille, Lille, France, for permission toreproduce the data shown as Fig. 1E and F.Duality of Interest. No potential conflicts ofinterest relevant to this article were reported.Author Contributions. M.G.W., J.A.M.S., andR.T. each researched published and personaldata, jointly synthesized ideas, and wrote themanuscript.

References1. Rudenski AS, Hadden DR, Atkinson AB, et al.Natural history of pancreatic islet B-cell functionin type 2 diabetesmellitus studied over six yearsby homeostasis model assessment. Diabet Med1988;5:36–412. Butler AE, Janson J, Bonner-Weir S, Ritzel R,Rizza RA, Butler PC. Beta-cell deficit and in-creased beta-cell apoptosis in humans withtype 2 diabetes. Diabetes 2003;52:102–1103. Franco M, Bilal U, Ordu~nez P, et al. Popula-tion-wide weight loss and regain in relation todiabetes burden and cardiovascular mortality inCuba 1980-2010: repeated cross sectional sur-veys and ecological comparison of seculartrends. BMJ 2013;346:f1515

Figure 3—During long-term excess calorie intake, especially in the presence of muscle insulinresistance, the raised plasma insulin levels will expedite chronic excess calorie storage fromcarbohydrate via de novo lipogenesis. This will promote storage of fat in the liver very graduallyover years and liver insulin resistance, with a consequent tendency for a small increase in plasmaglucose, as shown by theWhitehall II study (78). In turn, insulin secretion will increase to controlplasma glucose. The further increased insulin levels will bring about a self-reinforcingvicious cycle. The increased liver fat will inevitably lead to an increased rate of export of VLDL-triglyceride from the liver. Along with all other tissue, islets will therefore be exposed to higherrates of fatty acid supply, and the exposure of pancreatic endocrine cells to fatty acids and theirmetabolites will increase. This is postulated to bring about endoreticulum stress in susceptibleindividuals and eventually b-cell dedifferentiation, with relative inhibition of meal insulin secre-tion. The vicious cycles are postulated to interact overmany years. At a personal threshold level, itis postulated that the b-cells can no longer compensate, and plasma glucose levels will then riserelatively rapidly. The figure is redrawn from Taylor (75) and reproduced with permission fromTaylor (79).

2086 Reversible b-Cell Dysfunction Diabetes Care Volume 39, November 2016

4. Himsworth HP. Diet in the aetiology of hu-man diabetes. Proc R SocMed 1949;42:323–3265. Hrdlicka A, Ed. Physiological andMedical Ob-servations Among the Indians of SouthwesternUnited States and North Mexico. SmithsonianInstitution. Bureau of American Ethnology, Bul-letin 34. Washington, DC, Government PrintingOffice, 19086. Joslin EP. The universality of diabetes: a sur-vey of diabetes mortality in Arizona. J Am MedAssoc 1940;115:2033–20387. Knowler WC, Pettitt DJ, Savage PJ, BennettPH. Diabetes incidence in Pima indians: contri-butions of obesity and parental diabetes. Am JEpidemiol 1981;113:144–1568. Schulz LO, Bennett PH, Ravussin E, et al. Ef-fects of traditional and western environmentson prevalence of type 2 diabetes in Pima Indiansin Mexico and the U.S. Diabetes Care 2006;29:1866–18719. Jarrett RJ, Keen H, Fuller JH, McCartney M.Worsening to diabetes in men with impairedglucose tolerance (“borderline diabetes”). Dia-betologia 1979;16:25–3010. Taylor R. Aetiology of non-insulin depen-dent diabetes. Br Med Bull 1989;45:73–9111. Leslie RD, Pyke DA. Genetics of Diabetes. InDiabetes Annual. Alberti KG, Krall LP, Eds. Am-sterdam, Elsevier, 1985, p. 53–6612. Rosenbaum S, Skinner RK, Knight IB, GarrowJS. A survey of heights and weights of adults inGreat Britain, 1980. Ann Hum Biol 1985;12:115–12713. Hu FB, Manson JE, Stampfer MJ, et al. Diet,lifestyle, and the risk of type 2 diabetes mellitusin women. N Engl J Med 2001;345:790–79714. UKPDS. UK Prospective Diabetes Study(UKPDS). VIII. Study design, progress and per-formance. Diabetologia 1991;34:877–89015. Gregg EW, Cheng YJ, Narayan KM, ThompsonTJ, Williamson DF. The relative contributions ofdifferent levels of overweight and obesity to theincreased prevalence of diabetes in the UnitedStates: 1976-2004. Prev Med 2007;45:348–35216. Ferrannini E, Nannipieri M, Williams K,Gonzales C, Haffner SM, Stern MP. Mode of on-set of type 2 diabetes from normal or impairedglucose tolerance. Diabetes 2004;53:160–16517. McCarthy MI, Zeggini E. Genome-wide as-sociation studies in type 2 diabetes. Curr DiabRep 2009;9:164–17118. Kahn SE. The importance of the beta-cell inthe pathogenesis of type 2 diabetes mellitus.Am J Med 2000;108(Suppl. 6a):2S–8S19. UK Prospective Diabetes Study (UKPDS)Group. Effect of intensive blood-glucose controlwith metformin on complications in overweightpatients with type 2 diabetes (UKPDS 34). Lan-cet 1998;352:854–86520. Yang L, Ji W, Xue Y, Chen L. Imaging beta-cell mass and function in situ and in vivo. J MolMed (Berl) 2013;91:929–93821. Akirav EM, Lebastchi J, Galvan EM, et al.Detection of b cell death in diabetes using dif-ferentially methylated circulating DNA. ProcNatl Acad Sci U S A 2011;108:19018–1902322. Rahier J, Guiot Y, Goebbels RM, Sempoux C,Henquin JC. Pancreatic b-cell mass in Europeansubjects with type 2 diabetes. Diabetes ObesMetab 2008;10(Suppl. 4):32–4223. Lee Y, Hirose H, Ohneda M, Johnson JH,McGarry JD, Unger RH. Beta-cell lipotoxicity in

the pathogenesis of non-insulin-dependent di-abetes mellitus of obese rats: impairment inadipocyte-beta-cell relationships. Proc NatlAcad Sci U S A 1994;91:10878–1088224. Elks ML. Chronic perifusion of rat islets withpalmitate suppresses glucose-stimulated insulinrelease. Endocrinology 1993;133:208–21425. Lalloyer F, Vandewalle B, Percevault F, et al.Peroxisome proliferator-activated receptor al-pha improves pancreatic adaptation to insulinresistance in obese mice and reduces lipotoxic-ity in human islets. Diabetes 2006;55:1605–161326. Zhou YP, Grill VE. Long-term exposure of ratpancreatic islets to fatty acids inhibits glucose-induced insulin secretion and biosynthesisthrough a glucose fatty acid cycle. J Clin Invest1994;93:870–87627. Pinnick K, Neville M, Clark A, Fielding B.Reversibility of metabolic and morphologicalchanges associated with chronic exposure ofpancreatic islet beta-cells to fatty acids. J CellBiochem 2010;109:683–69228. Marchetti P, Bugliani M, Lupi R, et al. Theendoplasmic reticulum in pancreatic beta cellsof type 2 diabetes patients. Diabetologia 2007;50:2486–249429. Huang CJ, Lin CY, Haataja L, et al. High ex-pression rates of human islet amyloid polypep-tide induce endoplasmic reticulum stressmediated beta-cell apoptosis, a characteristicof humans with type 2 but not type 1 diabetes.Diabetes 2007;56:2016–202730. Laybutt DR, Preston AM, Akerfeldt MC,et al. Endoplasmic reticulum stress contributesto beta cell apoptosis in type 2 diabetes. Diabe-tologia 2007;50:752–76331. Poitout V, Amyot J, Semache M, Zarrouki B,Hagman D, Fontes G. Glucolipotoxicity of thepancreatic beta cell. Biochim Biophys Acta2010;1801:289–29832. Ferner RE, Ashworth L, Tronier B, Alberti KGEffects of short-term hyperglycemia on insulinsecretion in normal humans. Am J Physiol 1986;250:E655–E66133. Kramer CK, Choi H, Zinman B, Retnakaran R.Determinants of reversibility of b-cell dysfunc-tion in response to short-term intensive insulintherapy in patients with early type 2 diabetes.Am J Physiol Endocrinol Metab 2013;305:E1398–E140734. Carpentier AC, Bourbonnais A, Frisch F,Giacca A, Lewis GF. Plasma nonesterified fattyacid intolerance and hyperglycemia are associ-ated with intravenous lipid-induced impairmentof insulin sensitivity and disposition index. J ClinEndocrinol Metab 2010;95:1256–126435. Storgaard H, Jensen CB, Vaag AA, Vølund A,Madsbad S. Insulin secretion after short- andlong-term low-grade free fatty acid infusion inmen with increased risk of developing type 2diabetes. Metabolism 2003;52:885–89436. Lim EL, Hollingsworth KG, Aribisala BS, ChenMJ, Mathers JC, Taylor R. Reversal of type 2 di-abetes: normalisation of beta cell function in as-sociation with decreased pancreas and livertriacylglycerol. Diabetologia 2011;54:2506–251437. Steven S, Hollingsworth KG, Al-Mrabeh A,et al. Very low-calorie diet and 6 months ofweight stability in type 2 diabetes: pathophysi-ological changes in responders and nonre-sponders. Diabetes Care 2016;39: 808–815

38. Steven S, Hollingsworth KG, Small PK, et al.Weight loss decreases excess pancreatic triacyl-glycerol specifically in type 2 diabetes. DiabetesCare 2016;39:158–16539. Brereton MF, Iberl M, Shimomura K, et al.Reversible changes in pancreatic islet structureand function produced by elevated blood glu-cose. Nat Commun 2014;5:463940. Spijker HS, Song H, Ellenbroek JH, et al. Lossof b-cell identity occurs in type 2 diabetes and isassociated with islet amyloid deposits. Diabetes2015;64:2928–293841. Talchai C, Xuan S, Lin HV, Sussel L, Accili D.Pancreaticb cell dedifferentiation as amechanismof diabetic b cell failure. Cell 2012;150:1223–123442. Wang Z, York NW, Nichols CG, Remedi MS.Pancreatic b cell dedifferentiation in diabetesand redifferentiation following insulin therapy.Cell Metab 2014;19:872–88243. White MG, Marshall HL, Rigby R, et al. Ex-pression of mesenchymal and a-cell phenotypicmarkers in islet b-cells in recently diagnoseddiabetes. Diabetes Care 2013;36:3818–382044. Weir GC, Aguayo-Mazzucato C, Bonner-Weir S. b-cell dedifferentiation in diabetes isimportant, but what is it? Islets 2013;5:233–23745. Jonas J-C, Sharma A, Hasenkamp W, et al.Chronic hyperglycemia triggers loss of pancre-atic b cell differentiation in an animal model ofdiabetes. J Biol Chem 1999;274:14112–1412146. Cnop M, Abdulkarim B, Bottu G, et al. RNAsequencing identifies dysregulation of the hu-man pancreatic islet transcriptome by the satu-rated fatty acid palmitate. Diabetes 2014;63:1978–199347. Kitamura YI, Kitamura T, Kruse J-P, et al.FoxO1 protects against pancreatic b cell failurethrough NeuroD and MafA induction. CellMetab 2005;2:153–16348. Gao T, McKenna B, Li C, et al. Pdx1 main-tains b cell identity and function by repressingan a cell program. Cell Metab 2014;19:259–27149. Marselli L, Suleiman M, Masini M, et al. Arewe overestimating the loss of beta cells in type 2diabetes? Diabetologia 2014;57:362–36550. Guo S, Dai C, Guo M, et al. Inactivation ofspecific b cell transcription factors in type 2 di-abetes. J Clin Invest 2013;123:3305–331651. Butler AE, Robertson RP, Hernandez R,Matveyenko AV, Gurlo T, Butler PC. Beta cellnuclear musculoaponeurotic fibrosarcoma on-cogene family A (MafA) is deficient in type 2diabetes. Diabetologia 2012;55:2985–298852. Cinti F, Bouchi R, Kim-Muller JY, et al. Evi-dence of b-cell dedifferentiation in humantype 2 diabetes. J Clin Endocrinol Metab 2016;101:1044–105453. Butler AE, Dhawan S, Hoang J, et al. b-celldeficit in obese type 2 diabetes, a minor role ofb-cell dedifferentiation and degranulation.J Clin Endocrinol Metab 2016;101:523–53254. Singhal P, Caumo A, Carey PE, Cobelli C,Taylor R. Regulation of endogenous glucoseproduction after a mixed meal in type 2 diabe-tes. Am J Physiol Endocrinol Metab 2002;283:E275–E28355. Unger RH. Glucagon physiology and patho-physiology in the light of new advances. Diabe-tologia 1985;28:574–57856. Henquin JC, Rahier J. Pancreatic alpha cellmass in European subjects with type 2 diabetes.Diabetologia 2011;54:1720–1725

care.diabetesjournals.org White, Shaw, and Taylor 2087

57. Yoon KH, Ko SH, Cho JH, et al. Selectivebeta-cell loss and alpha-cell expansion in pa-tients with type 2 diabetes mellitus in Korea.J Clin Endocrinol Metab 2003;88:2300–230858. Marroqui L, Masini M, Merino B, et al.Pancreatic a Cells are Resistant to MetabolicStress-induced Apoptosis in Type 2 Diabetes.EBioMedicine 2015;2:378–38559. Steven S, Carey PE, Small PK, Taylor R. Re-versal of type 2 diabetes after bariatric surgeryis determined by the degree of achieved weightloss in both short- and long-duration diabetes.Diabet Med 2015;32:47–5360. Schisler JC, Jensen PB, Taylor DG, et al. TheNkx6.1 homeodomain transcription factor sup-presses glucagon expression and regulates glu-cose-stimulated insulin secretion in islet betacells. ProcNatl Acad Sci U SA2005;102:7297–730261. Yoneda S, Uno S, Iwahashi H, et al. Predom-inance of b-cell neogenesis rather than replica-tion in humans with an impaired glucosetolerance and newly diagnosed diabetes. J ClinEndocrinol Metab 2013;98:2053–206162. Cnop M, Hughes SJ, Igoillo-Esteve M, et al.The long lifespan and low turnover of humanislet beta cells estimated by mathematical mod-elling of lipofuscin accumulation. Diabetologia2010;53:321–33063. Perl S, Kushner JA, Buchholz BA, et al. Sig-nificant human beta-cell turnover is limited tothe first three decades of life as determined byin vivo thymidine analog incorporation and ra-diocarbon dating. J Clin Endocrinol Metab 2010;95:E234–E239

64. Schrauwen-Hinderling VB, Kooi ME, HesselinkMK, et al. Impaired in vivo mitochondrial functionbut similar intramyocellular lipid content in pa-tients with type 2 diabetes mellitus and BMI-matched control subjects. Diabetologia 2007;50:113–12065. Lim EL, Hollingsworth KG, Smith FE,Thelwall PE, Taylor R. Inhibition of lipolysis intype 2 diabetes normalizes glucose disposalwithout change in muscle glycogen synthesisrates. Clin Sci (Lond) 2011;121:169–17766. Cruzat VF, Keane KN, Scheinpflug AL,Cordeiro R, Soares MJ, Newsholme P. Alanyl-glutamine improves pancreatic b-cell functionfollowing ex vivo inflammatory challenge. J En-docrinol 2015;224:261–27167. Steven S, Taylor R. Restoring normoglycae-mia by use of a very low calorie diet in longversus short duration type 2 diabetes. DiabetMed 2015;32:1149–115568. Hall TC, Pellen MG, Sedman PC, Jain PK.Preoperative factors predicting remission oftype 2 diabetes mellitus after Roux-en-Y gastricbypass surgery for obesity. Obes Surg 2010;20:1245–125069. Panunzi S, Carlsson L, De Gaetano A, et al.Determinants of diabetes remission and glyce-mic control after bariatric surgery. DiabetesCare 2016;39:166–17470. Bunck MC, Corner A, Eliasson B, et al. Ef-fects of exenatide on measures of b-cell func-tion after 3 years in metformin-treated patientswith type 2 diabetes. Diabetes Care 2011;34:2041–2047

71. Swinnen SG, Hoekstra JB, DeVries JH. Insu-lin therapy for type 2 diabetes. Diabetes Care2009;32(Suppl. 2):S253–S25972. Accili D, Ahren B, Boitard C, Cerasi E,Henquin JC, Seino S. What ails the b-cell? Di-abetes Obes Metab 2010;12(Suppl. 2):1–373. Kahn SE, Haffner SM, Heise MA, et al.;ADOPT Study Group. Glycemic durability of ro-siglitazone, metformin, or glyburide monother-apy. N Engl J Med 2006;355:2427–244374. Blum B, Roose AN, Barrandon O, et al. Re-versal ofb cell de-differentiation by a small mol-ecule inhibitor of the TGFb pathway. eLife 2014;3:e0280975. Taylor R. Pathogenesis of type 2 diabetes:tracing the reverse route from cure to cause.Diabetologia 2008;51:1781–178976. Logue J, Walker JJ, Leese G, et al.; ScottishDiabetes Research Network EpidemiologyGroup. Association between BMI measuredwithin a year after diagnosis of type 2 diabetesand mortality. Diabetes Care 2013;36:887–89377. Taylor R, Holman RR. Normal weight indi-viduals who develop type 2 diabetes: the per-sonal fat threshold. Clin Sci (Lond) 2015;128:405–41078. Tabak AG, Jokela M, Akbaraly TN, BrunnerEJ, Kivimaki M, Witte DR. Trajectories of glycae-mia, insulin sensitivity, and insulin secretion be-fore diagnosis of type 2 diabetes: an analysisfrom the Whitehall II study. Lancet 2009;373:2215–222179. Taylor R. Type 2 diabetes: etiology and re-versibility. Diabetes Care 2013;36:1047–1055

2088 Reversible b-Cell Dysfunction Diabetes Care Volume 39, November 2016