1638 Diabetes Care Volume 38, September 2015

COMMENTARY

Julio Rosenstock1 and Ele Ferrannini2

Euglycemic Diabetic Ketoacidosis:
A Predictable, Detectable, and
Preventable Safety Concern With
SGLT2 Inhibitors
Diabetes Care 2015;38:1638–1642 | DOI: 10.2337/dc15-1380

THE CASE AT HAND that as of May 2015 a total of 101 cases inhibitors, which could delay diagnosis
Recently, the U.S. Food and Drug Ad- of DKA have been reported worldwide in and treatment and even accelerate the
ministration (FDA) issued a Drug Safety EudraVigilance in T2D patients treated progressive metabolic deterioration. In-
Communication that warns of an in- with SGLT2 inhibitors, with an estimated terestingly, the large clinical development
creased risk of diabetic ketoacidosis exposure over 0.5 million patient-years. programs of the three marketed SGLT2
(DKA) with uncharacteristically mild to No clinical details were provided except inhibitors, comprising .40,000 T2D pa-
moderate glucose elevations (euglycemic for the mention that “all cases were seri- tients, bore no clear signal of DKA. Erondu
DKA [euDKA]) associated with the use ous and some required hospitalisation. Al- et al. (3), representing Janssen, the man-
of all the approved sodium–glucose though [DKA] is usually accompanied by ufacturer of canagliflozin, report a rela-
cotransporter 2 (SGLT2) inhibitors (1). high blood sugar levels, in a number of tively low frequency of DKA (15 cases,
This Communication was based on 20 these reports blood sugar levels were 12 on canagliflozin and 3 still blinded in
clinical cases requiring hospitalization only moderately increased” (2). the CANagliflozin cardioVascular Assess-
captured between March 2013 and With this background, it is very timely ment Study [CANVAS]) detected in a ret-
June 2014 in the FDA Adverse Event Re- that in this issue of Diabetes Care there rospective analysis of 17,596 participants
porting System database. The scarce are two articles on this subject. Erondu in the development program up to May
clinical data provided suggested that et al. (3) report cases of DKA in T2D 2015. The estimated incidence ratesd0.5,
most of the DKA cases were reported in from a large clinical development pro- 0.8, and 0.2 per 1,000 patient-years with
patients with type 2 diabetes (T2D), for gram and Peters et al. (4) discuss cases canagliflozin 100 mg, canagliflozin 300 mg,
whom this class of agents is indicated; from clinical practice observations of and comparator, respectivelydif under-
most likely, however, they were insulin- T1D and T2D patients. whelming, are double with the SGLT2
treated patients, some with type 1 dia- It is not unusual that serious safety inhibitor. Upon our inquiry, the other
betes (T1D). The FDA also identified issues related to a new drug go unde- two manufacturers of approved SGLT2
potential triggering factors such as inter- tected during the relatively short clinical inhibitors, AstraZeneca and Boehringer
current illness, reduced food and fluid development programs for regulatory Ingelheim, provided preliminary (unpub-
intake, reduced insulin doses, and his- drug approval. This is particularly true lished) figures that are even lower than the
tory of alcohol intake. The following when the safety issue is unexpected, oc- Janssen data. In more than 18,000 pa-
month, at the request of the European curring as an off-target effect, or only tients exposed to dapagliflozin in the ran-
Commission, the European Medicines emerges once the drug is used widely. If domized controlled T2D study program,
Agency (EMA) announced on 12 June serious enough, the issue may require a including DECLARE (Dapagliflozin Effect
2015 that the Pharmacovigilance Risk As- label warning and a mitigation plan or on Cardiovascular Events), the frequency
sessment Committee has started a review even consideration of drug withdrawal. of reported events suggestive of DKA
of all of the three approved SGLT2 inhib- DKA is an overt serious clinical condition (blinded and unblinded events) is less
itors (canagliflozin, dapagliflozin, and em- that may be missed only if presenting than 0.1%. Similarly in DECLARE, aim-
pagliflozin) to evaluate the risk of DKA in with mild to moderate hyperglycemia, ing for 17,150 patients randomized to
T2D (2). The EMA announcement claimed as it may be the case with use of SGLT2 dapagliflozin or placebo, the total number

1
Dallas Diabetes and Endocrine Center at Medical City, Dallas, TX
2
Institute of Clinical Physiology, Consiglio Nazionale delle Ricerche, Pisa, Italy
Corresponding author: Julio Rosenstock, juliorosenstock@dallasdiabetes.com.
© 2015 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit,
and the work is not altered.
See accompanying articles, pp. 1680 and 1687.
care.diabetesjournals.org Rosenstock and Ferrannini 1639

of reported blinded events of potential escalated back to an HbA1c of 8% in the of carbohydrate, possibly in conjunction
DKAs is less than 0.1% (E. Johnsson, posttrial years (8). In a recent report from with reduced insulin dose. The euDKA
AstraZeneca, personal communication). the T1D Exchange clinic registry (which reported in T2D patients with SGLT2 in-
In a retrospective analysis of randomized provides the best cross-sectional U.S. hibitor treatment has a different origin.
phase 2 and 3 empagliflozin trials (.13,000 data), the average HbA1c was ;8%, and Full-dose SGLT2 inhibition induces a
T2D participants), there were eight events only 30% achieved a goal HbA1c of ,7%, rapid increase in urinary glucose excre-
consistent with DKA with no imbalance severe hypoglycemia occurred in 9–20% tion, ranging 50–100 g/day equally in
observed between patients treated with of patients per year depending on age men and women and lasting slightly lon-
empagliflozin 10 mg (two events), empa- and diabetes duration, overweight/ ger than 24 h (14). In a typical 60-year-old,
gliflozin 25 mg (one event), and placebo obesity was present in 68% of patients, overweight T2D patient (BMI 28 kg/m2)
(five events). In the cardiovascular out- and, interestingly, DKA still occurred at consuming 50% of daily calories as
come trial EMPA-REG Outcome with 10% per year in patients aged 13–26 carbohydrate (15), this glucose loss
approximately 7,000 patients, the fre- years and at 4–5% per year in the older amounts to 17–34% of estimated carbo-
quency of reported blinded events of DKA patients (9). Therefore, it is not surpris- hydrate intake in men and 22–44% in
is less than 0.1% (U. Broedl, Boehringer ing that given the burden of T1D and its women. Of note, in a comparative study
Ingelheim, personal communication). challenging unmet needs, the pharma- of Japanese and European T2D patients
Of note, the canagliflozin data reported cological properties of SGLT2 inhibitors treated with an SGLT2 inhibitor, urinary
by Erondu et al. (3) appeared to have a prompted clinical development pro- glucose excretion was, if anything, larger
greater incidence of DKA, but 6 out of the grams seeking regulatory approval and in the former (averaging 110 g/day)
12 cases had evidence of latent autoim- attracted off-label use in T1D. The data than in the latter (60 g/day) groups (16);
mune diabetes in adults or T1D or tested presented in this issue of Diabetes Care thus, in the Japanese group (BMI 25 kg/m2),
positive for GAD65 antibodies, and, per- in T2D (3) and, in particular, the cases the glucose loss through the urine repre-
haps, some of the other cases may have associated with T1D as presented by sented 47% of estimated daily carbohy-
been T2D misdiagnoses. And even if the Peters et al. (4) do provide a good op- drate intake in men and 57% in women.
diagnosis was correct, most of the patients portunity to discuss how these agents In general, depending on body size, glo-
were on insulin treatment and were part of modulate the pathophysiology leading merular filtration rate, and degree of hy-
CANVAS, suggesting a more advanced to DKA. Thus, it is in this context that perglycemia, SGLT2-induced glucose loss
T2D stage with significant b-cell failure. we need to analyze the potential prob- can make up a substantial fraction of
The FDA did acknowledge that some lem of euDKA associated with SGLT2 in- daily carbohydrate availability.
of the cases occurred in T1D, where in- hibitors to provide a more realistic and Abstracting from a study in well-
creasing off-label use of SGLT2 inhibi- practical perspective. controlled drug-naı̈ve or metformin-
tors has been observed, most likely due treated T2D patients on chronic therapy
to the favorable insulin-independent THE PATHOPHYSIOLOGY with an SGLT2 inhibitor (17), plasma glu-
glucose-lowering and weight-loss effects. Ketosis results from restriction of carbo- cose levels decreased by 20–25 mg/dL
Indeed, preliminary proof-of-concept pi- hydrate usage with increased reliance both in the overnight fasted state and
lot studies in T1D have reported improve- on fat oxidation for energy production. following a mixed meal. As glucose is the
ments in short-term glucose control with The pathogenesis of DKA is well estab- chief stimulus for insulin release under all
less glucose variability, weight loss, and lished (10). Briefly, absolute insulin de- circumstances, plasma insulin levels also fell
lower insulin doses (5–7). Social media ficiency leads to reduced glucose (by ;10 pmol/L fasting and ;60 pmol/L
have disseminated initial favorable expe- utilization and enhanced lipolysis; in- postmeal). In contrast, plasma glucagon
riences and could have contributed to creased delivery of free fatty acids concentrations increased significantly,
the raised expectations, leading to (FFAs) to the liver coupled with raised partly because of a diminished paracrine
many T1D patients discussing with their glucagon levels promotes FFA oxidation inhibition by insulin (18) and possibly also
physicians the addition of an SGLT2 in- and production of ketone bodies. In because of decreased SGLT2-mediated
hibitor in an attempt to ameliorate their both T1D and T2D, DKA presents with glucose transport into a-cells (19). As a con-
diabetes control. Indeed, despite new in- marked hyperglycemia (.250 mg/dL, sequence, the calculated prehepatic insulin-
sulin analogs and technological improve- typically 350–800 mg/dL), profuse glycos- to-glucagon molar concentration ratio
ments in insulin delivery devices and uria (2–4 mg z min21 z kg21), and hyper- dropped from 9 to 7 mol/mol in the fasting
glucose monitoring systems, T1D re- ketonemia (plasma b-hydroxybutyrate state and from 29 to 24 mol/mol during
mains an intrusive and challenging dis- 4.2–11.0 mmol/L) (11,12). The hypergly- the meal. This hormonal shift, which re-
ease, fraught with wide glucose swings cemia of DKA is associated with extreme leases inhibition of gluconeogenesis in
and hypoglycemic episodes that frustrate insulin resistance, manifesting itself as the liver (20), augmented EGP both in
patients, families, and health care pro- markedly (.70%) reduced tissue glucose the fasting state and during the meal
viders. There is no better example than disposal and increased endogenous glu- (17). Insulin sensitivity, however, was
the Diabetes Control and Complications cose production (EGP) (12). improved, as also shown with the use
Trial (DCCT). Despite 6 years of monthly euDKA was originally defined as DKA of the euglycemic insulin clamp, as a re-
visits with outstanding diabetes treat- with plasma glucose levels ,300 mg/dL sult of attenuated glucotoxicity (21).
ment teams with limitless resources to occurring in young T1D patients, two- The difference in the pathophysiology
achieve an HbA1c of 7%, the T1D patients thirds of whom were female (13). The of DKA versus SGLT2 inhibitor–induced
in the intensive intervention group primary cause was reduced availability euDKA is schematized in Fig. 1. In euDKA,
1640 Commentary Diabetes Care Volume 38, September 2015

levels and predisposes to increased

ketogenesis.

THE CLINICAL LESSON

The evidence reviewed above suggests
that the risk of bona fide euDKA (and not
simple ketosis) in T2D related to the use
of SGLT2 inhibitors will probably turn
out to be very low, with an “acceptable”
frequency. Still, physicians and patients
need to be made aware that such risk
may be increased in long-standing T2D
patients with marked b-cell insuffi-
ciency or in latent autoimmune diabetes
in adults with rapid evolution toward
T1D and during prolonged starvation,
after surgery, or during intercurrent
illness. In T1D, however, the euDKA
risk appears to be more concrete for
Figure 1—Essential pathophysiology of DKA and euDKA consequent of the use of SGLT2 inhib- reasons entirely within the pathophysi-
itors. TGD, tissue glucose disposal; UGCr, urinary glucose clearance rate. ology: 1) in T1D patients hyperglycemia
typically is higher than in T2D patients,
2) in early T1D glomerular filtration rate
insulin deficiency and insulin resistance concentrations, nonoxidative glucose may be increased, 3) insulin may en-
are milder (and insulin resistance may disposal (i.e., glycogen synthesis and hance the effect of SGLT2 inhibition on
actually be improved); therefore, glu- lactate release) also fell by 15%. The glycosuria, and 4) changes in insulin
cose overproduction and underutiliza- augmented FFA delivery to the liver re- dose are not infrequent and may be in-
tion are quantitatively lesser than in sulted in mild stimulation of ketogene- appropriate for the amount and kind of
DKA. More importantly, renal glucose sis, whereby both fasting and mean carbohydrate intake.
clearance (i.e., the ratio of glycosuria postmeal b-hydroxybutyrate levels We submit that this potential compli-
to prevailing glycemia) is twice as large rose ;twofold higher than in the baseline cation related to SGLT2 inhibition is pre-
with euDKA than with DKA. In fact, study (though not exceeding 1 mmol/L); dictable, detectable, and preventable (or
from previous studies of patients ad- conversely, plasma lactate levels de- mitigable) so that the balance of benefits
mitted with DKA it can be calculated creased ;20%, a readout of reduced and risks favors the use of SGLT2 inhibi-
that renal glucose clearance averaged carbohydrate utilization (E. Ferrannini tors in the T1D population, which is in
0.3 mL z min21 z kg21 (12), whereas in et al., unpublished data). Had insulin desperate need of adjunct therapies. It
T2D patients it rose from a near- deficiency been more profounddas is predictable because the persistent gly-
negligible value in the baseline study can happen in T1D patientsdor had cosuria induced by SGLT2 inhibition sets
to 0.6 mL z min 21 z kg 21 with SGLT2 carbohydrate availability been drasti- off a sequence of metabolic changes that
treatment (17). Thus, it is the entity cally restricted, this mild ketosis would are obligatory quantitative consequences
of glycosuria viz. the height of hyper- have evolved toward ketoacidosis, with of a large glucose subtraction from the
glycemia that marks the difference be- decreased blood pH and bicarbonate body glucose pool. In particular, enhanced
tween the two metabolic states. This and increased anion gap (12). It must ketogenesis is seen already even in non-
difference can actually be amplified as be noted that in the only study that diabetic subjects receiving SGLT2 inhibi-
DKA frequently occurs in patients with looked at insulin administration (22)d tors and is inscribed on already modestly
impaired renal function (and hence less dating back to 1951dit was found that raised plasma b-hydroxybutyrate levels in
glycosuria) (11), while SGLT2 inhibitors insulin decreased the transport maxi- patients with diabetes (E. Ferrannini et al.,
may be used in patients with glomeru- mum for glucose in patients with diabe- unpublished observations). This back-
lar hyperfiltration (and more abundant tes, implying that starting or escalating ground ketonemia is asymptomatic and
glycosuria) (5). exogenous insulin treatment would in- clinically irrelevant in most T2D patients
Ketoacidosis follows with the same duce glycosuria in its own right. While but is certainly more of a concern in T1D
sequence of events in euDKA as in these studies have not been repeated patients who are already prone to de-
DKA. Thus, in SGLT2-treated T2D patients and the mechanism of this insulin effect velop ketosis under circumstances of
(17), the lower insulin-to-glucagon ratio has not been investigated, it is never- reduced insulin doses, stress, and inter-
stimulated lipolysis (circulating FFAs theless intriguing that insulin may in- current illnesses and following a hypo-
were ;40% higher during the meal) and tensify SGLT2-induced glycosuria. glycemic episode and during prolonged
enhanced lipid oxidation (by 20% on av- All in all, euDKA is pathophysiologi- fasting or starvation.
erage) at the expense of carbohydrate cally similar to DKA except for the Thus, one can envision a sequence of
oxidation (which fell by 60%). In the circumstancedSGLT2-induced glycosuriad clinical events that could evolve into a
face of lower substrate (glucose) that “artificially” lowers plasma glucose full-blown episode of DKA (Fig. 2). First,
care.diabetesjournals.org Rosenstock and Ferrannini 1641

the necessary information on the safety

and efficacy of SGLT2 inhibitors in T1D
(as well as insulin-treated T2D). Regula-
tory scrutiny of these compounds will
balance their beneficial impact on over-
all glycemic control, glycemic variability,
and weight management against the
risk of hypoglycemia and overall safety,
including risk of euDKA. A reduction in
insulin dose should not be regarded as a
positive outcome in itself and should be
achieved by slow, gentle decrements si-
multaneously to avoid hypoglycemia
and sliding toward euDKA. Hopefully,
these clinical development programs
will quickly expand so as to offer to pa-
tients and physicians a potential adjunct
against the day-to-day management
challenges of such a demanding disease.
Figure 2—Demonstration of the cascade of clinical events and metabolic changes that contrib-
ute sequentially to progressive clinical deterioration and development of full-blown episodes of
euDKA. BG, blood glucose; CHO, carbohydrate; TGs, triglycerides. Duality of Interest. J.R. has participated in
advisory boards and received honoraria or con-
sulting fees from Merck, Sanofi, Novo Nordisk, Eli
Lilly, MannKind Corporation, GlaxoSmithKline,
inappropriate reductions of insulin doses any time an SGLT2 inhibitor–treated pa- Takeda, Daiichi Sankyo, Boehringer Ingelheim,
or any factor that may increase insulin tient feels unwell regardless of the ambi- Janssen, Lexicon, and Intarcia and has received
demand, such as stress, a sick day, or ent glucose levels. This should be part of research grants from Merck, Pfizer, Sanofi, Novo
Nordisk, Eli Lilly, GlaxoSmithKline, Takeda, Novartis,
even alcohol intake, may induce hyper- any educational element for those treated
AstraZeneca, Janssen, Daiichi Sankyo, MannKind
ketonemia. Under these circumstances, with an SGLT2 inhibitor. If detectable, Corporation, Bristol-Myers Squibb, Boehringer
initially patients may just not feel well then euDKA is preventable because de- Ingelheim, Lexicon, and Intarcia. E.F. has been a mem-
or experience some malaise and perhaps tection of significant ketonuria and/or ber of the scientific advisory board or a speaker
mild nausea with no vomiting. Their first ketonemia any time symptoms such as for Boehringer Ingelheim, Merck, AstraZeneca,
Eli Lilly, Johnson & Johnson, Takeda, Novo Nordisk,
impulse is to check their blood glucose; nausea and/or vomitingdor even just GlaxoSmithKline, and Sanofi and has received re-
because of the persistent glycosuria, malaisedappear, especially after alcohol search grant support from Eli Lilly and Boehringer
glycemia will be only mildly elevated so intake or a recent cut in insulin dose, can Ingelheim. No other potential conflicts of inter-
that they would tend to reduce or with- prompt advice to maintain vigorous fluid est relevant to this article were reported.
hold insulin and avoid eating. These ma- intake and to consume carbohydrates to
References
neuvers will accelerate ketone production allow at least full-dose insulin therapy un-
1. U.S. Food and Drug Administration. Drug
and metabolic decompensation toward til the ketosis resolves. Patients should Safety Communication: FDA warns that SGLT2 in-
DKA. The metabolic picture will be further temporarily stop the SGLT2 inhibitor, con- hibitors for diabetes may result in a serious con-
compounded by the volume depletion tact their medical provider, and take sup- dition of too much acid in the blood [Internet], 15
caused by the persistent glycosuria and plemental boluses of rapid insulin along May 2015. Available from http://www.fda.gov/
vomiting. with liquids and carbohydrates. Even if downloads/Drugs/DrugSafety/UCM446954.pdf.
Accessed 22 June 2015
The cases reported by Peters et al. (4) patients are unable to adjust the insulin 2. European Medicines Agency. Review of dia-
exemplify some of the factors that trig- dose, euDKA can be mitigable by drinking betes medicines called SGLT2 inhibitors started:
gered DKA: most commonly they were and eating as tolerated without fear of risk of diabetic ketoacidosis to be examined [In-
insulin reductions, low caloric and fluid hyperglycemia and seeking prompt med- ternet], 12 June 2015. Available from http://
www.ema.europa.eu/docs/en_GB/document_
intake, intercurrent illness, and alcohol ical attention for parenteral fluid replace-
library/Referrals_document/SGLT2_inhibitors__20/
use. Time was wasted because of delayed ment and insulin therapy. Procedure_started/WC500187926.pdf. Accessed 22
diagnosis due to deceptively “acceptable” In any event, T1D patients who choose June 2015
blood glucose levels. Peters et al. (4) do to take this medication off-label should 3. Erondu N, Desai M, Ways K, Meininger G.
provide a further understanding of the sign an ad hoc informed consent that Diabetic ketoacidosis and related events in the
canagliflozin type 2 diabetes clinical program.
clinical clues that can contribute to the makes them fully aware of the potential Diabetes Care 2015;38:1680–1686
early detection and help raise awareness for euDKA, the precipitating factors, the 4. Peters AL, Buschur EO, Buse JB, Cohan P,
of the potential for SGLT2 inhibition to warning symptoms and signs, and the Diner JC, Hirsch IB. Euglycemic diabetic ketoaci-
cause euDKA in T1D. We believe that preventative measures to adopt. dosis: a potential complication of treatment
euDKA is in fact easily detectable because with sodium–glucose cotransporter 2 inhibi-
reliable tools are currently available to CONCLUSIONS tion. Diabetes Care 2015;38:1687–1693
5. Perkins BA, Cherney DZ, Partridge H, et al.
monitor ketonuria and ketonemia and The ongoing long-term, randomized, Sodium-glucose cotransporter 2 inhibition and
should be recommended to be used at placebo-controlled studies will provide glycemic control in type 1 diabetes: results of an
1642 Commentary Diabetes Care Volume 38, September 2015

8-week open-label proof-of-concept trial. Dia- KGMM, Eds. New York, John Wiley & Sons, Ltd., 17. Ferrannini E, Muscelli E, Frascerra S, et al.
betes Care 2014;37:1480–1483 2015, p. 799–814 Metabolic response to sodium-glucose cotrans-
6. Henry RR, Rosenstock J, Edelman S, et al. Ex- 11. Barrett EJ, DeFronzo RA, Bevilacqua S, porter 2 inhibition in type 2 diabetic patients. J
ploring the potential of the SGLT2 inhibitor Ferrannini E. Insulin resistance in diabetic keto- Clin Invest 2014;124:499–508
dapagliflozin in type 1 diabetes: a randomized, acidosis. Diabetes 1982;31:923–928 18. Maruyama H, Hisatomi A, Orci L, Grodsky
double-blind, placebo-controlled pilot study. 12. Luzi L, Barrett EJ, Groop LC, Ferrannini E, GM, Unger RH. Insulin within islets is a physio-
Diabetes Care 2015;38:412–419 DeFronzo RA. Metabolic effects of low-dose in- logic glucagon release inhibitor. J Clin Invest
7. Sands AT, Zambrowicz BP, Rosenstock J, sulin therapy on glucose metabolism in diabetic 1984;74:2296–2299
et al. Sotagliflozin, a dual SGLT1 and SGLT2 ketoacidosis. Diabetes 1988;37:1470–1477 19. Bonner C, Kerr-Conte J, Gmyr V, et al. In-
inhibitor as adjunct therapy to insulin in type 13. Munro JF, Campbell IW, McCuish AC, hibition of the glucose transporter SGLT2 with
1 diabetes. Diabetes Care 2015;38:1181– Duncan LJP. Euglycaemic diabetic ketoacidosis. dapagliflozin in pancreatic alpha cells triggers
1188 BMJ 1973;2:578–580 glucagon secretion. Nat Med 2015;21:512–
8. Nathan DM; DCCT/EDIC Research Group. The 14. Sha S, Devineni D, Ghosh A, et al. 517
Diabetes Control and Complications Trial/ Canagliflozin, a novel inhibitor of sodium glucose 20. Burgess SC. Regulation of glucose metabo-
Epidemiology of Diabetes Interventions and Com- co-transporter 2, dose dependently reduces cal- lism in liver. In International Textbook of Diabe-
plications study at 30 years: overview. Diabetes culated renal threshold for glucose excretion and tes Mellitus. 4th ed. DeFronzo RA, Ferrannini E,
Care 2014;37:9–16 increases urinary glucose excretion in healthy Zimmet P, Alberti KGMM, Eds. New York, John
9. Miller KM, Foster NC, Beck RW, et al.; T1D subjects. Diabetes Obes Metab 2011;13:669–672 Wiley & Sons, Ltd., 2015, p. 193–210
Exchange Clinic Network. current state of type 1 15. Hall KD, Chow CC. Estimating changes in 21. Merovci A, Solis-Herrera C, Daniele G, et al.
diabetes treatment in the U.S.: updated data free-living energy intake and its confidence in- Dapagliflozin improves muscle insulin sensitiv-
from the T1D Exchange clinic registry. Diabetes terval. Am J Clin Nutr 2011;94:66–74 ity but enhances endogenous glucose produc-
Care 2015;38:971–978 16. Ferrannini E, Veltkamp SA, Smulders RA, tion. J Clin Invest 2014;124:509–514
10. Kitabchi AE, Umpierrez GE, Murphy MB. Di- Kadokura T. Renal glucose handling: impact of 22. Farber SJ, Berger EY, Earle DP. Effect of di-
abetic ketoacidosis and hyperosmolar state. In chronic kidney disease and sodium-glucose co- abetes and insulin of the maximum capacity of
International Textbook of Diabetes Mellitus. 4th transporter 2 inhibition in patients with type 2 the renal tubules to reabsorb glucose. J Clin In-
ed. DeFronzo RA, Ferrannini E, Zimmet P, Alberti diabetes. Diabetes Care 2013;36:1260–1265 vest 1951;30:125–129