TODAY Study ORIGINAL ARTICLE

Safety and Tolerability of the Treatment of Youth-Onset Type 2 Diabetes The TODAY experience

TODAY STUDY GROUP* safety and efficacy of three distinct ap- proaches to treatment of youth-onset type 2 diabetes. The rationale, design, and meth- OBJECTIVEdData related to the safety and tolerability of treatments for pediatric type 2 ods (4), baseline characteristics of the cohort diabetes are limited. The TODAY clinical trial assessed severe adverse events (SAEs) and targeted (5), and primary outcomes (POs) (6) have nonsevere adverse events (AEs) before and after treatment failure, which was the primary out- been reported previously. The purpose of come (PO). this article is to report the safety, tolerability, RESEARCH DESIGN AND METHODSdObese 10- to 17-year-olds (N = 699) with type and event rates of the interventions used in 2 diabetes for ,2 years and hemoglobin A1c (A1C) #8% on metformin monotherapy were TODAY, paying particular attention to the randomized to one of three treatments: metformin, metformin plus rosiglitazone (M + R), or side effects of metformin (gastrointestinal metformin plus lifestyle program (M + L). Participants were followed for 2–6.5 years. [GI] symptoms, anemia, lactic acidosis) and rosiglitazone (hepatotoxicity, anemia, RESULTSdGastrointestinal (GI) disturbance was the most common AE (41%) and was lower in P edema, weight gain, heart failure, fractures) the M + R group ( = 0.018). Other common AEs included anemia (20% before PO, 14% after PO), previously reported in adults. abnormal liver transaminases (16, 15%), excessive weight gain (7, 9%), and psychological events (10, 18%); the AEs were similar across treatments. Permanent medication reductions/discontinuations occurred most often because of abnormal liver transaminases and were lowest in the M + R group RESEARCH DESIGN AND (P = 0.005). Treatment-emergent SAEs were uncommon and similar across treatments. Most METHODS (98%) were unrelated or unlikely related to the study intervention. There were no deaths and only 18 targeted SAEs (diabetic ketoacidosis, n = 12; severe hypoglycemia, n = 5; lactic acidosis, n =1). There were 62 pregnancies occurring in 45 participants, and 6 infants had congenital anomalies. Study design The TODAY study design and methods CONCLUSIONSdThe TODAY study represents extensive experience managing type 2 di- have been reported previously (4). abetes in youth and found that the three treatment approaches were generally safe and well Briefly, TODAY was a multicenter, ran- tolerated. Adding rosiglitazone to metformin may reduce GI side effects and hepatotoxicity. domized, three-arm, parallel-group clini- – cal trial funded by the National Institute Diabetes Care 36:1765 1771, 2013 of Diabetes and Digestive and Kidney Dis- eases (NIDDK) of the National Institutes iabetes is increasing in prevalence U.S. (3), and by 2001 this had risen to of Health. The collaborative study group Dworldwide and results in consider- 22–76% (2), depending on race/ethnicity included 15 clinical centers and a data co- able morbidity and mortality at an and geographic location. Despite the ordinating center (these centers are listed estimated annual cost of more than $174 growing prevalence of youth-onset type in the Supplementary Data). Participants billion in the U.S. alone (1). Although 2 diabetes, long-term experience to define who met eligibility criteria at the end of typically a disease of adults, over the last the efficacy and safety of treatment ap- run-in were randomized (1:1:1) to one of two decades type 2 diabetes has become proaches is limited. Only metformin and three groups: 1) metformin alone (1,000 an important pediatric disorder. On the insulin currently are approved by the mg twice daily) (M); 2) metformin plus basis of data from the SEARCH for Diabe- Food and Drug Administration for treat- rosiglitazone (4 mg twice daily) (M + R); tes in Youth Study Group (2) and a review ment of type 2 diabetes in children. or 3) metformin plus an intensive lifestyle by Fagot-Campagna et al. (3), by the mid- The Treatment Options for type 2 Dia- intervention (M + L) (7). Assignment to 1990s, type 2 diabetes accounted for betes in Adolescents and Youth (TODAY) the M or M + R groups was double- 16–45% of youth with diabetes in the clinical trial was designed to address the blinded. Participants were recruited over 4.5 years and followed for 2–6.5 years. ccccccccccccccccccccccccccccccccccccccccccccccccc The PO of TODAY was time to treat- Corresponding author: Laura Pyle, [email protected]. ment failure, defined as either 1)hemo- Received 15 November 2012 and accepted 11 March 2013. globin A1c (A1C) $8% sustained for 6 DOI: 10.2337/dc12-2390. Clinical trial reg. no. NCT00081328, clinicaltrials.gov. months or 2) the inability to wean from This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10 .2337/dc12-2390/-/DC1. insulin within 3 months after acute met- The opinions expressed in this article are those of the authors and do not necessarily reflect the views of the abolic decompensation. After reaching respective Tribal and Indian Health Service Institution Review Boards or their members. PO, rosiglitazone was discontinued, met- *A complete list of the members of the TODAY Study Group can be found in the Supplementary Data online. formin was continued, and insulin glar- The members of the writing group are listed in the APPENDIX. gine was started and titrated to achieve © 2013 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 http://creativecommons.org/ theglycemictarget. licenses/by-nc-nd/3.0/ for details. The protocol was approved by an See accompanying articles, pp. 1732, 1735, 1742, 1749, 1758, 1772, 1775, and external evaluation committee convened 1777. by the NIDDK and by the institutional care.diabetesjournals.org DIABETES CARE, VOLUME 36, JUNE 2013 1765 Safety of type 2 diabetes treatments in TODAY review board for the protection of human tabulation of rates of AEs associated with at baseline, and economic status, was an- subjects of each participating institution. study interventions. Targeted nonserious alyzed separately before and after PO or All participants provided informed con- AEs included clinically manifest heart fail- censoring (i.e., study termination without sent and minor children confirmed assent ure, anemia, renal impairment, excessive PO). Analyses included all randomized according to local guidelines. An indepen- weight gain (defined as a .10% increase participants in their assigned treatment dent data and safety monitoring board in BMI between visits), edema, psycholog- groups. The study was powered for the (DSMB) convened by NIDDK reviewed ical events, recurrent mild hypoglycemia, PO only; the secondary outcomes repor- progress, safety, and interim analyses through- GI symptoms, and mildly (1.5–2.5 ted here are considered exploratory. P , out the study. 3ULN) or definitively elevated (.2.5 0.05 is considered statistically significant. 3ULN) AST, ALT, or both. Study sample Serious adverse events.SAEswereclas- RESULTSdCharacteristics of the ran- Eligibility criteria included age 10–17 sified as 1)death,2) life threatening med- domized cohort overall and by group are years (inclusive), type 2 diabetes based ical event, 3)birthofababywitha shown in Table 1. Participants were on American Diabetes Association criteria congenital anomaly, 4) hospitalization or 14.0 6 2.0 years old and had diabetes (8) for ,2 years at randomization, BMI prolongation of hospitalization, 5)dis- for 7.8 6 5.8 months at the time of ran- higher than the 85th percentile for age ability, 6) overdose of study medication, domization. Nearly two-thirds were and sex, and the ability to complete a or 7) event requiring intervention to pre- women; about 80% were from racial/ethnic 2- to 6-month run-in period taking met- vent an SAE. In addition, the study spec- minority groups (32.5% non-Hispanic formin monotherapy with A1C ,8%. Ex- ified the following SAEs of particular black, 39.7% Hispanic, 5.9% American clusion criteria relevant to safety included interest: severe hypoglycemia, diabetic Indian, 1.6% Asian) and 20.3% were creatinine clearance ,70 mL/min/1.73 m2; ketoacidosis (DKA), and lactic acidosis. non-Hispanic white. There were no statis- hepatic transaminase (aspartate amino- Causation of SAEs was assigned by the tical differences for any baseline variable transaminase [AST] or alanine amino- investigator as not, unlikely, possibly, or by treatment group. transferase [ALT]) .2.5 times the upper probably related to study participation limit of normal (3ULN); hematocrit (Hct) or interventions. Using the SAE Tracker, Targeted adverse events ,30% or hemoglobin (Hgb) ,10 g/dL; or each SAE was reviewed by the TODAY The number of participants experiencing women who were pregnant, planning SMC (see below). targeted AEs by treatment group assign- pregnancy, or failed to practice appropri- Safety monitoring. An SMC consisting ment, before and after reaching PO, is ate contraception. of study group members held biweekly shown in Table 2. Forty percent reported A total of 699 participants were ran- conference calls to review events and GI symptoms before PO and 44% after domized and monitored every 2 months other issues related to safety. SMC mem- PO. Before PO, fewer participants repor- in the first year and quarterly thereafter. bers were masked to treatment group ted GI symptoms in the M + R group (33 Study visits included physical examination assignments and the chair reported any vs. 41% in the M group and 45% in the (including blood pressure, assessment of concerns to the Committee for Oversight M+Lgroup;P = 0.0054). Only five par- peripheral edema, anthropometrics); labo- of Procedures and the steering committee. ticipants required protocol-driven medi- ratory assessments (A1C, Hgb/Hct, AST/ Any event category that occurred more cation dose reductions because of GI ALT, serum creatinine, vitamin B12); and often than a predetermined frequency symptoms. The only two AEs that oc- assessments of interim adverse events was referred to the Committee for Over- curred with a frequency exceeding the (AEs). sight of Procedures to consider a need for predetermined thresholds for concern further referral to the independent DSMB. (3% in both cases) were definitively ele- Recording and reviewing of adverse Before participant enrollment, the DSMB vated liver transaminases (.2.5 3ULN) events developed criteria for interim assessment and excessive weight gain (defined as a The TODAY clinical trial assessed serious of safety and treatment success or futility. .10% increase in BMI between visits). AEs (SAEs) and targeted nonserious AEs The DSMB met twice yearly (and as After exceeding the threshold, these before and after PO. The Supplementary needed) for unmasked assessments. events were tracked in an unblinded man- Data gives definitions and study-specific ner by the DSMB, which identified no responses for targeted AEs and definitions Statistical methods concerns. Definitively elevated liver trans- for study-specificSAEs. Descriptive statistics are reported as me- aminases occurred in 5.7% of the partic- Adverse events.Targetednonserious dian, mean 6 SD, or percentage. The ipants before PO and 5.0% after PO and AEs were recorded using an electronic number of participants experiencing AEs were not significantly different among online Safety and Comorbidity Tracker and SAEs were analyzed using logistic re- treatment groups. Mildly elevated trans- developed for TODAY and were reviewed gression (SAS PROC GENMOD, version aminases (1.5–2.5 3ULN) occurred in regularly by the safety and monitoring 9.2; SAS Institute Inc., Cary, NC). Rates 10% before PO and 10% after PO. Before committee (SMC) that was blinded to and counts of AEs and SAEs were ana- PO there was no difference among treat- treatment group. The purposes of track- lyzed using zero-inflated Poisson regres- ment groups, but after PO there was mar- ing AEs were to 1) identify treatment- sion (SAS PROC GENMOD). To examine ginal significance across groups for mildly related AEs and facilitate adherence to whether any AEs of treatments were al- elevated transaminases (P =0.048).Ex- study-specified guidelines and algo- tered by discontinuation of rosiglitazone cessive weight gain occurred in 6.9% be- rithms, 2) determine whether AEs were and initiation of insulin therapy (Lantus, fore PO and 8.8% after PO. There was no occurring at a higher than anticipated over- Sanofi-Aventis, Bridgewater, NJ) at the statistically significant difference among all rate or a disproportionate rate at one time of PO, the effect of treatment group, treatment groups for our defined AE of center compared with others, and 3)permit after adjusting for sex, race/ethnicity, age excessive weight gain.

1766 DIABETES CARE, VOLUME 36, JUNE 2013 care.diabetesjournals.org TODAY Study Group

Table 1dBaseline characteristics overall and by treatment group PO showed a difference between groups (1.26 in the M, 0.97 in the M + R, 0.33 in the M + L groups; P = 0.029), with only Overall Metformin M+R M+L fi Characteristics (N = 699) (n =232) (n =233) (n =234) the signi cant difference that between M+LandMgroups. Age (years) 14.0 6 2.0 14.1 6 1.9 14.1 6 2.1 13.8 6 2.0 Although rosiglitazone has been re- Duration of diabetes (months) 7.8 6 5.8 7.8 6 6.0 8.0 6 5.7 7.6 6 5.8 ported to increase the risk of bone frac- Female sex (%) 64.9 63.1 65.7 66.0 tures (9), after controlling for age, sex, BMI (kg/m2) 34.9 6 7.6 35.8 6 8.1 35.0 6 7.7 34.1 6 7.1 race/ethnicity, and socioeconomic status BMI z score 2.23 6 0.47 2.27 6 0.45 2.22 6 0.49 2.19 6 0.48 (household income and education at A1C (%) 7.1 6 2.2 7.3 6 2.2 7.0 6 2.3 7.1 6 2.2 baseline), there was no difference in frac- Hemoglobin (mg/dL) 13.4 6 1.6 13.4 6 1.6 13.4 6 1.6 13.4 6 1.5 ture rate among treatment groups. Hematocrit (%) 39.6 6 4.0 39.6 6 4.1 39.7 6 4.1 39.6 6 3.9 AST/SGOT (U/L) 26.9 6 21.1 28.5 6 23.8 25.8 6 20.4 26.2 6 18.7 Serious adverse events ALT/SGPT (U/L) 34.0 6 33.7 37.1 6 34.4 30.5 6 29.3 34.1 6 36.5 Table 4 shows SAEs overall and by study Creatinine clearance (mL/min) 158 6 37 162 6 36 157 6 40 156 6 35 group assignment before and after PO; 42 Data are presented as mean 6 SD unless otherwise indicated. SGOT, serum glutamic-oxaloacetic trans- events were recorded in more than one aminase; SGPT, serum glutamic-pyruvic transaminase. category. Most categories contained in- sufficient numbers of events to be statis- tically meaningful, and there were no Anemia occurred in 20% of partici- experienced repeated mild hypoglycemia differences among the groups in the cate- pants before PO and 14% after PO. There in the M + L group (0.9%) compared with gories for which analysis was possible. was no difference in the percentage of M + R (3.3%; P = 0.020) and M (3.3%; P = Hospitalization was responsible for 91 participants experiencing anemia among 0.013) groups. Except for psychological and 94% of the SAEs before and after treatment groups before PO, but after PO events, which occurred in 10% before and PO, respectively. The vast majority of more participants in the M group experi- 18% after PO, the remaining AEs were SAEs (89%) were not related to study par- enced anemia than in the M + R group (19 infrequent, such that an assessment of be- ticipation or intervention and 98% were and 10%, respectively; P = 0.014). The tween group differences was not possible. either not or unlikely related. Only two occurrence of repeated mild hypoglyce- Table 3 shows the number of targeted SAEs were deemed probably related and mia was low (3.9% before and 2.5% AEs (per 100 participant-years of expo- two possibly related. Of these four events, after PO). Before PO, the percentage of sure) by treatment group assignment be- two were severe hypoglycemia, one was participants experiencing repeated mild fore and after reaching PO and represents hypoglycemia resulting in hospitaliza- hypoglycemia was similar between 1,782 participant-years of exposure be- tion, and one was an episode of DKA oc- groups. However, when the M + R group fore PO and 876 participant-years after curring in a participant for whom study was compared with the other two (non- PO. In Table 3, participants could be re- medication had been discontinued for rosiglitazone) groups combined, more corded as experiencing multiple episodes safety reasons (definitively elevated transa- participants taking rosiglitazone (6.9%) of an event if the event resolved and then minases; see below). There were no partic- experienced repeated mild hypoglycemia recurred. It should be noted that upon ipant deaths during the TODAY study. than those not taking rosiglitazone (2.4%; reaching PO, rosiglitazone was discontin- There were 12 cases of DKA (in 11 partic- P = 0.03). After PO, fewer participants ued in all participants. Only edema before ipants) and 5 occurrences (in 5 participants)

Table 2dParticipants with targeted adverse events overall and by treatment group

Before primary outcome After primary outcome Overall M M+R M+L Overall M M+R M+L AEs (N = 699) (n =232) (n =233) (n = 234) P* (N = 319) (n =120) (n =90) (n =109) P* Heart failure 0 0 0 0 d 1010d Anemia 140 50 50 40 0.3795 45 23 9 13 0.0367a Renal impairment 2 0 1 1 d 0000d LFTs 1.5–2.5 3 ULN732721250.64963217870.0482a .2.5 3 ULN 40 18 9 13 0.1260 16 4 2 10 0.0906 Edema 93 4 20.47295113d Excessive weight gain 48 10 21 17 0.3592 28 6 13 9 0.4665 Psychological 69 19 25 25 0.4885 58 22 12 24 0.3016 Repeated mild hypoglycemia 27 5 16 6 0.0939 8 4 3 1 0.0286b,c GI symptoms 280 96 78 106 0.0181c 139 55 36 48 0.2753

LFT, liver function test. *Event categories with no P value recorded had too few events to be statistically analyzed. aM vs. M + R significant. bMvs.M+Lsignificant. cM+Rvs.M+Lsignificant. care.diabetesjournals.org DIABETES CARE, VOLUME 36, JUNE 2013 1767 Safety of type 2 diabetes treatments in TODAY

Table 3dNumber of targeted adverse events per 100 participant-years of exposure overall and by treatment group

Before primary outcome After primary outcome AEs Overall M M + R M + L P* Overall M M + R M + L P* Participant-years of exposure (per 100 years) 17.82 5.56 6.18 6.08 8.76 3.40 2.43 2.93 Heart failure 0.00 0.00 0.00 0.00 d 0.11 0.00 0.41 0.00 d Anemia 11.11 12.59 11.65 9.20 0.6130 6.28 7.95 4.12 6.13 d Renal impairment 0.11 0.00 0.16 0.16 d 0.00 0.00 0.00 0.00 d LFTs 1.5–2.5 3 ULN 4.66 5.94 3.72 4.44 0.7186 4.34 6.18 4.12 2.39 d .2.5 3 ULN 2.58 3.96 1.78 2.14 0.1289 2.05 1.18 0.82 4.09 d Edema 0.84 1.26 0.97 0.33 0.0285a 0.80 0.59 0.41 1.36 d Excessive weight gain 2.75 1.80 3.40 2.96 0.4801 3.54 1.77 6.59 3.07 d Psychological 4.71 4.50 4.85 4.77 0.9745 8.22 7.36 7.82 9.54 0.3958 Repeated mild hypoglycemia 1.96 1.26 3.40 1.15 0.1528 0.91 1.18 1.24 0.34 d GI symptoms 26.03 27.35 22.82 28.10 0.4172 26.60 25.61 26.77 27.60 0.4715

LFT, liver function test. *Event categories with no P value recorded had too few events to be statistically analyzed. aMvs.M+Lsignificant. of severe hypoglycemia. Despite a study re- acidosis (lactate level 9.8 mmol/L) occurred because of definitively elevated liver trans- quirement for appropriate contraceptive in a 15-year-old girl assigned to the M group aminases (AST, 211 IU/L; ALT, 256 IU/L). measures while taking study medication, during a severe asthma exacerbation. Aside there were 62 pregnancies in 45 partici- from asthma, she was asymptomatic and Discontinuations of study medication pants during the TODAY study, and 6 the elevated lactate resolved within 12 h and primary outcome (10%) were associated with a congenital of asthma treatment. This was assessed by The TODAY protocol specified blinded anomaly. For those planning pregnancy, the local investigator and the SMC and de- dose reduction/discontinuation in re- study medication was discontinued before termined to be unlikely related to metfor- sponse to certain targeted AEs (defini- attempts at conception. For those young min use. A single event of congestive heart tively elevated transaminases, severe women who became unexpectedly preg- failure occurred 6 days postpartum after a hypoglycemia, intercurrent illness, nant, study medication was discontinued pregnancy complicated by hypertension. edema, anemia, and renal impairment). at the first indication of pregnancy. Although this occurred in a participant in Overall, 85 AEs occurring in 67 partici- Pregnancy tests were performed on all the M + R group, the subject already had pants (9.6%) resulted in protocol-driven female participants at every visit. The 14 been off rosiglitazone for approximately discontinuation of study medication or life-threatening SAEs were not significantly 2 years 9 months before the heart failure de- permanent dose reduction, consistent different among groups; of these, 7 were veloped. Therefore, this event was classified with protocol-driven safety algorithms; psychiatric hospitalizations due to suicidal as not related to study intervention or par- 65% of these medication alterations (55 ideation/attempt, 2 were DKA, 1 was severe ticipation. No life-threatening SAEs were of 85 occurring in 42 participants) were hyperglycemia and ketosis (without acido- classified as probably related, and the only due to elevated liver transaminases. Ad- sis), 2 were infections (one associated with a one classified as possibly related to study justment because of transaminase eleva- deep vein thrombosis, the other a postpar- participation was an episode of DKA occur- tion was significantly different among tum infection), and 1 was lactic acidosis ring in a participant for whom study med- treatment groups (28 in the M group, 8 in associated with asthma. This case of lactic ication was discontinued 2 months earlier the M + R group, 19 in the M + L group; P =

Table 4dSerious adverse events overall and by treatment group

Before primary outcome After primary outcome Overall M M+R M+L Overall M M+R M+L SAEs (n =119) (n =28) (n =33) (n =58) (n =144) (n =49) (n =46) (n = 49) Lifethreatening 8413 6150 Hospitalization 108 24 32 52 136 47 45 44 Disability 1 0 0 1 0 0 0 0 Requires intervention to prevent SAE 15 6 2 7 12 3 3 6 Congenital abnormality* 2 1 0 1 4 1 0 3 Event from overdose of study medication 0 0 0 0 2 0 0 2 Lacticacidosis 1100 0000 Severe hypoglycemia 4 0 1 3 1 1 0 0 DKA 3210 9423

Rows below do not sum to total because each event could have more than one summary characteristic checked. *Of 62 pregnancies in 45 participants.

1768 DIABETES CARE, VOLUME 36, JUNE 2013 care.diabetesjournals.org TODAY Study Group

0.005). The remaining events were of such however, in the normal range. The prev- (17) to 0.35 after 24 weeks (15), which low frequency that no group differences alence of anemia among women was more compares to about 0.4 at 12 weeks and could be detected. A post hoc analysis of than twice that of men (32 vs. 15%; P , 0.7 at 24 weeks among participants in the effect of certain presumed side effects 0.0001). TODAY’s M + R group (6). Weight gain of rosiglitazone (edema, elevated transa- from rosiglitazone seems to be a potential minases) upon medication dose reduc- CONCLUSIONSdOur data demon- problem in children and adolescents, as it tions showed that the rate of dose strate that the three treatment approaches is in adults. alterations due to liver transaminase eleva- are all generally safe and well tolerated by Nonalcoholic fatty liver disease tions differed in the M + R group. A lower adolescents. Other than a high rate of (NAFLD) is the most common cause of rate was observed in the participants trea- reported GI disturbances (41%), there elevated liver transaminases in the U.S. ted with rosiglitazone than in those not were few treatment-related AEs overall and is strongly associated with type 2 taking rosiglitazone (0.93 vs. 2.61 per pa- and little difference in the frequency of diabetes (19). Because of hepatotoxicity tient-year; P , 0.005). Finally, those who AEs and SAEs among treatment groups. associated with troglitazone, we were par- had discontinued permanently or reduced The high rate of GI events is not un- ticularly concerned about potential he- study medication did not reach PO sooner expected because all participants were patic effects of rosiglitazone. Although than those who did not. taking metformin, which has been asso- liver transaminase elevations were com- ciated with GI symptoms in more than a mon, there were no significant differences fi Anemia and plasma vitamin B12 quarter of children (12) and adults (13) in in the event rates of mildly or de nitively concentrations clinical trials of shorter duration. The elevated liver transaminases among the Anemia was diagnosed in 20% of partic- prevalence of GI events in the M + R group treatment groups either before or after ipants before PO, whereas 14% had either (33%) was significantly lower than in the PO (Table 3). Therefore, based on 1,782 an initial diagnosis or a reoccurrence after M + L (45%) or the combined nonrosigli- patient-years of experience during the PO. Before PO, cases of anemia occurred tazone (43%) groups. The reason for this TODAY study, there is no evidence of almost equally among all three treatment difference is unclear, but it cannot be at- hepatotoxicity related to rosiglitazone in groups. After PO, there were more anemia tributed to any disparity in metformin youth with type 2 diabetes. Furthermore, events in the M group compared with the dose or medication compliance, which the need for permanent dosage reduc- M + R group. Only a single participant in were similar among the three treatment tions and medication discontinuations the M + L group required a permanent arms (6). due to elevated liver transaminases was reduction in medication dose because of It is notable that the M + R group did significantly lower in the M + R group. anemia. Since all participants in this trial not have more AEs than the other The etiology of the elevated trans- were taking metformin, which can impair two groups. Peripheral edema and aminases was not formally evaluated as absorption of vitamin B12 (10,11) and excessive weight gain are known adverse part of this study. Nonetheless, partici- contribute to anemia, we examined effects of rosiglitazone as well as other pants with abnormal AST or ALT levels plasma concentrations of vitamin B12 in thiazolidinediones (14–18). However, underwent an evaluation for other causes all participants. Additional tests to iden- peripheral edema occurred in only 1.7% of liver disease. In the preponderance of tify the causes of anemia (such as iron of participants in the M + R group, which cases, no genetic, infectious, or autoim- studies) were performed at the discretion was not significantly more than in the mune cause of elevated transaminases was of local clinic staff, but these data were not other two groups and compares favorably identified. Therefore, we suspect that the collected for analysis. Median vitamin B12 to the 4–6% occurrence of edema in sim- high rate of abnormal liver transaminases concentrations were similar across treat- ilar trials of adults (17,18). In addition, reflects a high prevalence of NAFLD in ment groups at baseline and 2 years, and using the protocol-defined AE criteria these obese pediatric participants with the proportion of participants with vita- set a priori (.10% increase in BMI type 2 diabetes. The prevalence of min B12 concentrations in the deficient between visits), excessive weight gain NAFLD in obese children is estimated to (,203 pg/mL), borderline (203–299 pg/ occurred in 9.0% of the M + R group com- be 9.6% (20), and this incidence is esti- mL), and normal ($300 pg/mL) ranges paredwith4.3and7.3%intheMand mated to approach 50% in obese youth were also similar among treatment M + L groups, respectively (not statisti- with type 2 diabetes (21). Because hepatic groups. There was no meaningful change cally significant). The study has reported steatosis is strongly associated with insu- in vitamin B12 concentrations over time. gains in BMI in the M + R group, which lin resistance (22), the high incidence of When classified as deficient, borderline, increased significantly more over 60 elevated liver transaminases in the or normal, lower vitamin B12 concentra- study months than BMI in either the M TODAY study likely reflects the magni- tions were significantly associated with or M + L groups (6). The apparent contra- tude of hepatic insulin resistance in chil- anemia. At 2,695 participant visits during diction between equivalent incidence of dren with type 2 diabetes. Examining the which both Hgb/Hct and vitamin B12 excessive weight gain between 3-month effect of insulin-sensitizing agents on were assessed, 315 participants had ane- visits across treatment and significantly NAFLD, Nadeau et al. (21) reported that mia and, of these, 100 were associated higher BMI across 60 months in the M + R five of six children with type 2 diabetes with deficient or borderline vitamin B12 group can be explained by the specificity showed an improvement in liver transa- values. Those study participants who had of the study’sdefinition of excessive minases after treatment. We postulate anemia tended to have lower vitamin B12 weight gain as an AE, which was designed that the trend toward decreased fre- concentrations at the visit at which ane- primarily to provide an early signal of safety quency of transaminase abnormalities mia was detected than those without ane- concerns. Studies of overweight adults trea- and the reduced number of medication mia (median 355 with vs. 408 without; ted with rosiglitazone have shown a gain in dose adjustments in the M + R group P , 0.0001). These median values were, BMI ranging from 1.1 kg/m2 after 12 weeks may be the result of improved insulin care.diabetesjournals.org DIABETES CARE, VOLUME 36, JUNE 2013 1769 Safety of type 2 diabetes treatments in TODAY sensitivity (23). Similar effects of rosigli- agents with diverse mechanisms of action. UL1-RR025780 (to University of Colorado tazone have been described in adults Although the treatment options used in Denver). (14,24). the TODAY study were safe and well tol- N.H.W. is on data monitoring committees Although anemia was commonly ob- erated, the safety and efficacy of additional for Novo Nordisk and Daiichi Sankyo. M.W.H. served over the course of this study, it was agents in adolescents must be evaluated is on advisory panels for Novo Nordisk and Daiichi Sankyo, on a data safety monitoring distributed equally across treatment swiftly to reduce the growing number of committee for Bristol-Myers Squibb, and on a groups. Metformin can impair vitamin young adults at risk for the devastating scientific advisory committee for Xeris Phar- B12 absorption and presumably increase consequences of poorly controlled type 2 maceuticals. No other potential conflicts of the risk of megaloblastic anemia (10,11). diabetes. interest relevant to this article were reported. Since all treatment groups in the TODAY N.H.W., S.M.W., M.W.H., and K.J.N. re- study were taking metformin, we cannot searched data, contributed to the discussion, assess the effects of metformin on vitamin APPENDIXd wrote the manuscript, and reviewed and edi- The members of the ted the manuscript. L.P., R.G., and D.E.H. B12 and anemia in this study. Although writing group are as follows: Neil H. researched data, contributed to the discussion, our data demonstrate that lower vitamin White (chair), MD, CDE, Department of B concentrations were associated with and reviewed/edited the manuscript. T.P. re- 12 Pediatrics and Medicine, Washington searched data and reviewed and edited the anemia, the majority of participants with fi UniversityinSt.Louis;LauraPyle,PhD, manuscript. S.D.C. and S.N. contributed to anemia did not exhibit vitamin B12 de - George Washington University Bio- the discussion and reviewed and edited the ciency. Since anemia was more than twice statistics Center; Steven M. Willi, MD, manuscript. L.P. is the guarantor of this work as prevalent in young women as in young Department of Pediatrics, Children’sHos- and, as such, had full access to all the data in men, we speculate that the principal cause the study and takes responsibility for the in- fi pital of Philadelphia; Trang Pham, MS, of anemia in this group was iron de ciency MPH, George Washington University Bio- tegrity of the data and the accuracy of the data analysis. secondary to menstrual blood loss. statistics Center; Steven D. Chernausek, We were reassured to confirm a very The TODAY Study Group thanks the fol- MD, Department of Pediatrics, University of lowing companies for donations in support of low incidence of renal impairment; mount- Oklahoma Health Sciences Center; Robin ’ ing evidence suggests that renal com- the study s efforts: Becton, Dickinson and Goland, MD, Naomi Berrie Diabetes Center, Company; Bristol-Myers Squibb; Eli Lilly plications (macroalbuminuria, dialysis, Columbia University; Daniel E. Hale, MD, and Company; GlaxoSmithKline; LifeScan; transplantation) in youth-onset type 2 di- fi fi fi Department of Pediatrics, University of P zer; and Sano . The TODAY Study Group abetes can manifest within the rst 10 years Texas Health Science Center at San Antonio; gratefully acknowledges the participation and of disease (25,26). Severe hypoglycemia Morey W. Haymond, MD, Department of guidance of the American Indian partners as- also occurred infrequently, as might be ex- Pediatrics, Baylor College of Medicine; sociated with the clinical center located at the pected from the treatments used in the KristenJ.Nadeau,MD,DepartmentofPe- University of Oklahoma Health Sciences Center, including members of the Absentee TODAY study, which rarely predispose pa- diatrics, University of Colorado Denver; tients to hypoglycemia in the absence of Shawnee Tribe, Cherokee Nation, Chickasaw and Sumana Narasimhan, MD, Depart- Nation, Choctaw Nation of Oklahoma, and exogenous insulin therapy. Nearly 1 in 10 ment of Pediatrics, Case Western Re- pregnancies resulted in congenital malfor- Oklahoma City Area Indian Health Service. serve University. Materials developed and used for the mation; this is nearly twice the frequency TODAY standard diabetes education pro- reported in a recent meta-analysis (27). gram and the intensive lifestyle intervention Finally, the high rate of psychological AEs AcknowledgmentsdThe TODAY clinical program are available to the public at https:// in the TODAY study is consistent with a trial was funded by the National Institute of today.bsc.gwu.edu/. previously reported 19% prevalence of Diabetes and Digestive and Kidney Diseases neuropsychiatric diseases in children at (NIDDK) of the National Institutes of Health the time of diagnosis of type 2 diabetes (28). (NIH). This work was completed with funding References Type 2 diabetes is a serious medical from the NIDDK/NIH grant numbers U01- 1. Centers for Disease Control and Pre- condition in adolescents, with high mor- DK61212, U01-DK61230, U01-DK61239, vention. National diabetes fact sheet: na- bidity and limited treatment options. The U01-DK61242, and U01-DK61254; the Na- tional estimates and general information PO data from the TODAY study showed tional Center for Research Resources General on diabetes and prediabetes in the United fi Clinical Research Centers Program grant States, 2011. Atlanta, GA, U.S. Depart- that M + R had the greatest ef cacy in numbers M01-RR00043-45 (to Children’s ment of health and Human Services, maintaining glycemic control, suggesting Hospital Los Angeles), M01-RR00069 (to Uni- Centers for Disease Control and Preven- that aggressive action against hypergly- versity of Colorado Denver), M01-RR00084 (to tion. Available from http://www.cdc.gov/ cemia in the early stages of the disease the Children’s Hospital of Pittsburgh), M01- diabetes/pubs/pdf/ndfs_2011.pdf. Accessed using a combination of pharmacologic RR01066 (to Massachusetts General Hospital), 18 October 2012 approaches may be warranted (6). Al- M01-RR00125 (to Yale University), and M01- 2. Liese AD, D’Agostino RB Jr, Hamman RF, though rosiglitazone is no longer widely RR14467 (to University of Oklahoma Health et al.; SEARCH for Diabetes in Youth available because of concerns of increased Sciences Center); and the National Center for Study Group. The burden of diabetes risk of myocardial infarction (29), the TO- Research Resources Clinical and Translational mellitus among US youth: prevalence es- DAY experience demonstrates that com- Science Awards grant numbers UL1-RR024134 timates from the SEARCH for Diabetes in (to Children’s Hospital of Philadelphia), UL1- Youth Study. Pediatrics 2006;118:1510– bining rosiglitazone with metformin RR024139 (to Yale University), UL1-RR024153 1518 provides greater durability of glycemic ’ fi (to Children sHospitalofPittsburgh),UL1- 3. Fagot-Campagna A, Pettitt DJ, Engelgau control without a signi cant increase in RR024989 (to Case Western Reserve Uni- MM, et al. Type 2 diabetes among North AEs. The spectrum of antidiabetic agents versity), UL1-RR024992 (to Washington American children and adolescents: an to emerge since the inception of the University, St. Louis), UL1-RR025758 epidemiologic review and a public health TODAY study is remarkable and includes (to Massachusetts General Hospital), and perspective. J Pediatr 2000;136:664–672

1770 DIABETES CARE, VOLUME 36, JUNE 2013 care.diabetesjournals.org TODAY Study Group

4. Zeitler P, Epstein L, Grey M, et al.; TODAY 12. Jones KL, Arslanian S, Peterokova VA, of fatty liver in children and adolescents. Study Group. Treatment options for type Park JS, Tomlinson MJ. Effect of metfor- Pediatrics 2006;118:1388–1393 2 diabetes in adolescents and youth: a study min in pediatric patients with type 2 21. Nadeau KJ, Klingensmith G, Zeitler P. of the comparative efficacy of metformin diabetes: a randomized controlled trial. Type 2 diabetes in children is frequently alone or in combination with rosiglitazone Diabetes Care 2002;25:89–94 associated with elevated alanine amino- or lifestyle intervention in adolescents with 13. DeFronzo RA, Goodman AM; The Multi- transferase. J Pediatr Gastroenterol Nutr type 2 diabetes. Pediatr Diabetes 2007;8: center Metformin Study Group. Efficacy 2005;41:94–98 74–87 of metformin in patients with non-insulin- 22. Wicklow BA, Wittmeier KD, MacIntosh 5. Copeland KC, Zeitler P, Geffner M, et al.; dependent diabetes mellitus. N Engl J Med AC, et al. Metabolic consequences of hepatic TODAY Study Group. Characteristics of 1995;333:541–549 steatosis in overweight and obese adoles- adolescents and youth with recent-onset 14. Yki-Järvinen H. Thiazolidinediones. N Engl cents. Diabetes Care 2012;35:905–910 type 2 diabetes: the TODAY cohort at J Med 2004;351:1106–1118 23. TODAY Study Group. Effects of metfor- baseline. J Clin Endocrinol Metab 2011; 15. Khanolkar MP, Morris RHK, Thomas AW, min, metformin plus rosiglitazone, and 96:159–167 et al. Rosiglitazone produces a greater re- metformin plus lifestyle on insulin sensi- 6. Zeitler P, Hirst K, Pyle L, et al.; TODAY duction in circulating platelet activity tivity and b-cell function in TODAY. Di- Study Group. A clinical trial to maintain compared with gliclazide in patients with abetes Care 2013;36:1749–1757 glycemic control in youth with type 2 di- type 2 diabetes mellitusdan effect prob- 24. Yki-Järvinen H. Thiazolidinediones and abetes. N Engl J Med 2012;366:2247–2256 ably mediated by direct platelet PPAR- the liver in humans. Curr Opin Lipidol 7. TODAY Study Group. Design of a family- gamma activation. Atherosclerosis 2008; 2009;20:477–483 based lifestyle intervention for youth with 197:718–724 25. Dart AB, Sellers EA, Martens PJ, Rigatto C, type 2 diabetes: the TODAY study. Int J 16. Kim YM, Cha BS, Kim DJ, et al. Predictive Brownell MD, Dean HJ. High burden of Obes (Lond) 2010;34:217–226 clinical parameters for therapeutic efficacy kidney disease in youth-onset type 2 di- 8. American Diabetes Association. Diagnosis of rosiglitazone in Korean type 2 diabetes abetes. Diabetes Care 2012;35:1265–1271 and classification of diabetes mellitus. Di- mellitus. Diabetes Res Clin Pract 2005;67: 26. Krakoff J, Lindsay RS, Looker HC, Nelson abetes Care 2005;28(Suppl. 1):S37–S42 43–52 RG, Hanson RL, Knowler WC. Incidence 9.KahnSE,ZinmanB,LachinJM,etal.;Di- 17. Berberoglu Z, Yazici AC, Bayraktar N, of retinopathy and nephropathy in youth- abetes Outcome Progression Trial (ADOPT) Demirag NG. Rosiglitazone decreases onset compared with adult-onset type 2 Study Group. Rosiglitazone-associated frac- fasting plasma peptide YY3-36 in type 2 diabetes. Diabetes Care 2003;26:76–81 tures in type 2 diabetes: an analysis from diabetic women: a possible role in weight 27. Balsells M, Garcıa-Patterson A, Gich I, A Diabetes Outcome Progression Trial gain? Acta Diabetol 2012;49(Suppl. 1): Corcoy R. Maternal and fetal outcome in (ADOPT). Diabetes Care 2008;31:845–851 S115–S122 women with type 2 versus type 1 diabetes 10. de Jager J, Kooy A, Lehert P, et al. Long 18. Nesto RW, Bell D, Bonow RO, et al. mellitus: a systematic review and meta- term treatment with metformin in patients Thiazolidinedione use, fluid retention, analysis. J Clin Endocrinol Metab 2009;94: with type 2 diabetes and risk of vitamin and congestive heart failure: a consensus 4284–4291 B-12 deficiency: randomised placebo statement from the American Heart Asso- 28. Levitt Katz LE, Swami S, Abraham M, et al. controlled trial. BMJ 2010;340:c2181 ciation and American Diabetes Associa- Neuropsychiatric disorders at the pre- 11. Reinstatler L, Qi YP, Williamson RS, tion. Diabetes Care 2004;27:256–263 sentation of type 2 diabetes mellitus in Garn JV, Oakley GP Jr. Association of 19. Clark JM, Brancati FL, Diehl AM. The children. Pediatr Diabetes 2005;6:84–89 biochemical B12 deficiency with metformin prevalence and etiology of elevated ami- 29. Nissen SE, Wolski K. Rosiglitazone re- therapy and vitamin B12 supplements: the notransferase levels in the United States. visited: an updated meta-analysis of risk National Health and Nutrition Examination Am J Gastroenterol 2003;98:960–967 for myocardial infarction and cardiovas- Survey, 1999-2006. Diabetes Care 2012; 20. Schwimmer JB, Deutsch R, Kahen T, cular mortality. Arch Intern Med 2010; 35:327–333 Lavine JE, Stanley C, Behling C. Prevalence 170:1191–1201

care.diabetesjournals.org DIABETES CARE, VOLUME 36, JUNE 2013 1771