Lipid and Inflammatory Cardiovascular Risk Worsens Over 3 Years in Youth With Type 2 Diabetes: The TODAY clinical trial

The study design has been reported in detail (8,9,11). Briefly, the collaborative study group included 15 clinical centers, a data-coordinating center, and a central laboratory (see Supplementary Data). At baseline (July 2004–February 2009), eligible individuals were 10–17 years of age with type 2 diabetes diagnosed <2 years at time of randomization and with a BMI ≥85th percentile. During a prerandomization 2–6-month run-in period, all participants received standard diabetes education. After randomization, participants attended clinic visits for purposes of medical management every 2 months in the first year and quarterly thereafter; outcome data were collected at baseline, 6 months, and annually.

The lifestyle program was designed to work with pharmacotherapy to improve diabetes control in youth through sustained, moderate weight loss (7–10% of initial body weight or the equivalent for youth still growing in height). The program, as previously described in detail (9), consisted of family-based behavioral change delivered in a series of in-person visits during the first 2 years, followed by continued contact at quarterly medical visits. Primary behavior-change targets included energy-balance behaviors (dietary and physical activity) and family involvement/support. A trained interventionist delivered the program one-on-one to the youth and family using materials specifically developed for TODAY (e.g., manuals, booklets, logs, and fact sheets).

Lipid goals were defined in the study protocol as LDL <100 mg/dL and triglycerides <150 mg/dL. If lipid levels were outside the target range, initial therapy involved dietary counseling. If LDL levels remained ≥130 mg/dL or if triglyceride levels remained 300–599 mg/dL after 6 months of nutrition and diabetes management, pharmacological treatment with atorvastatin was initiated and adjusted to achieve target goals according to an algorithm based on lipid levels (8). If triglycerides were ≥600 mg/dL, fibrate therapy could be initiated at the discretion of the physician. Additional cardiovascular risk factors, including LDL particle density, apoB, NEFA, hsCRP, PAI-1, and homocysteine levels, were measured throughout the TODAY trial.

The protocol was approved by an External Evaluation Committee convened by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health and by the Institutional Review Boards for the Protection of Human Subjects of each participating institution. All participants provided informed consent, and minor children confirmed assent according to local guidelines. A Data and Safety Monitoring Board convened by the National Institute of Diabetes and Digestive and Kidney Diseases reviewed progress, safety, and interim analyses throughout the study.

Measurements of fasting total plasma cholesterol, cholesterol in the lipoprotein fractions, and triglycerides were performed enzymatically on a Roche Modular P autoanalyzer (Roche Diagnostics, Indianapolis, IN), standardized to the Centers for Disease Control and Prevention Reference Methods. HDL was measured after precipitation of apoB-containing particles by dextran sulfate Mg²⁺. LDL was calculated by the Friedewald equation (12), except if triglyceride levels were >400 mg/dL, in which case ultracentrifugation was performed using the Lipid Research Clinic Beta Quantification procedure. Non-HDL was defined as the difference between total cholesterol and HDL. The interassay coefficients of variation (CVs) were <1.5% for triglycerides and <2% for HDL. LDL physical properties and the flotation distribution of other lipoprotein fractions were determined by a single ultracentrifugation procedure (13). LDL buoyancy (relative flotation rate [Rf]) was calculated as the LDL peak fraction divided by the total number of fractions collected. The Rf of each plasma sample calculated by this procedure is highly reproducible with CV <1.8%. Small dense LDL was defined as an Rf of ≤0.263. Immunochemical determination of apoB concentration in plasma was performed on the Siemens Nephelometer BN II (Siemens Healthcare Diagnostics Inc., Newark, DE) using Siemens reagents and in-house–prepared calibrator and quality-control materials.

Levels of hsCRP in plasma were measured immunochemically using Siemens reagents (Siemens Healthcare Diagnostics Inc.) on a Siemens nephelometer autoanalyzer (BNII). Homocysteine levels were measured in plasma samples by enzymatic assay with reagents from Axis Shield Inc. (Bothell, WA) on a Roche Modular P analyzer (Roche Diagnostics). Measurement of plasma PAI-1 levels was performed using a quantikine ELISA kit from R&D Systems (Minneapolis, MN). The assay sensitivity was 0.15 ng/mL, and it was linear up to 20 ng/mL. The interassay CVs for the high, medium, and low PAI-1 controls were 6.1, 6.4, and 9.5%, respectively. Analysis of NEFA was performed using reagents from Wako Diagnostics (Richmond, VA) on a Roche Hitachi Modular P analyzer (Roche Diagnostics).

HbA_1c concentration was measured with a dedicated high-performance liquid chromatography method (TOSOH Biosciences Inc., South San Francisco, CA), certified by the National Glycohemoglobin Standardization Program.

Longitudinal data were analyzed using generalized linear mixed models to account for the multiple observations per participant (SAS PROC MIXED and PROC GENMOD, version 9.2; SAS Institute Inc., Cary, NC). Standard cutoffs for risk categories were applied to six of the outcomes (LDL, triglycerides, HDL by sex, hsCRP, apoB, and LDL particle size). Analyses included all data available at each of the four annual visit time points (baseline and 12, 24, and 36 months), as long as prior to attainment of the primary outcome. Variables not normally distributed were log-transformed for testing; descriptive summary statistics and plots are presented using the original scale. HDL was analyzed separately for males and females due to the known differential in levels. Effects of treatment group, visit (time), sex, and race-ethnicity (American Indian and Asian combined into "other" category due to sample size) and their interactions were tested.

Each lipid outcome was regressed on HbA_1c with treatment group and the treatment-by-HbA_1c interaction in the model; the effect of BMI was determined by adding BMI as a covariate adjustment. This analysis included follow-up data only (i.e., months 12, 24, and 36) and removed values collected while the participant was taking a lipid-lowering medication.

The P values <0.05 are identified as statistically significant with no adjustment for multiple testing. The TODAY study was powered for the primary time-to-failure outcome only, although secondary analyses were predefined.

At baseline, participants were 14.0 (± SD 2.0) years of age, obese (BMI 34.9 ± 7.6 kg/m²), with a mean duration of type 2 diabetes of 7.8 (± 5.9) months. Participants were all pubertal, and 88.7% were Tanner stage 4 or 5. The majority (64.7%) were female, 32.5% were non-Hispanic black (NHB), 39.7% Hispanic (H), 20.3% non-Hispanic white (NHW), 5.9% American Indian, and 1.6% Asian. Patient flow through the study is presented in the CONSORT diagram (10) (Supplementary Fig. 1).

Prevalence of high-risk LDL (i.e., LDL ≥130 mg/dL or taking LDL-lowering medications) increased from 4.5 to 10.7% over 36 months, while the proportion of subjects at target (LDL <100 mg/dL) declined (Table 1). At baseline, 6 participants used lipid-lowering drugs (4 in M, 2 in M+L); by 36 months, an additional 47 participants were using lipid-lowering drugs (46 on statin, 1 on sequestrant). Treatment-group assignment was not statistically different between those who did and did not use lipid-lowering drugs. Demographics of those treated were not different from the cohort as a whole. Of the 46 treated with statin, 32 had at least one follow-up lipid panel prior to reaching primary outcome, of whom only 10 achieved LDL goal.

For all variables described below, Table 1 shows longitudinal trends in mean values from baseline to 36 months; Table 2 shows trends for percent of the cohort with risk factor values above the threshold for high risk. Longitudinal descriptive statistics for all variables discussed, with stratification by treatment group, sex, and race/ethnicity, are provided in more detail in the Supplementary Material (Supplementary Tables 1 and 2).

LDL levels rose from baseline to month 12 and remained at this higher level at months 24 and 36 (P < 0.0001) (Fig. 1A). Of the 517 participants who attended any of the three annual follow-up visits, 74 (14.3%) had at least one high-risk classification (LDL ≥130 mg/dL or taking lipid-lowering medication), and only 289 (55.9%) were always at goal (LDL <100 mg/dL). A similar pattern was observed for apoB and non-HDL (P < 0.0001; Table 1).

Triglyceride values rose significantly from baseline to month 12 (P = 0.0038) and remained at the higher level over months 24 and 36. In males, HDL rose from baseline to month 12 and then maintained this higher level (P < 0.0001). Females also followed this trend but also showed more variability over follow-up (month 12 > 24, P = 0.0075; month 24 < 36, P = 0.0437). The percentage of small dense LDL was significantly higher at baseline than at months 12, 24, and 36 (P = 0.0043); as small dense LDL percentage usually increases as triglycerides increase, this trend was unexpected.

A greater percentage of participants had high-risk hsCRP (>0.3 mg/dL) at baseline in the M compared with the M+R group (P = 0.0136; Table 2). Overall, in the entire cohort, there was a slight increase over time in percent of participants with high-risk hsCRP levels (41.2% at baseline and 46.3% at month 36; P = 0.0217). Homocysteine increased significantly over time (P < 0.0001), jumping from baseline to month 12, leveling off to month 24, and jumping again to month 36, while PAI-1 values rose significantly (P = 0.0059) from baseline to month 12 and remained at the higher level thereafter (Table 1). There were significant differences in mean NEFA over time (P < 0.0001), with a drop from baseline to 12 months, followed by an increase at 24 months and then a leveling off from month 24 to 36 (Table 1).

In Fig. 1, vertical axes are scaled to distinguish lipid values, apoB, and hsCRP by treatment group. Longitudinal analysis of triglycerides found a significant difference between M and M+L (P = 0.0035), with lower values associated with intensive lifestyle (Fig. 1D and Supplementary Table 1B). For LDL (Fig. 1A), apoB (Fig. 1B), HDL (Fig. 1C), and non-HDL (not shown), there were no statistically significant differences among treatment groups over the 36 months. Small dense LDL was less common at all time points in M+R (P = 0.0001) and M+L (P = 0.0121) in comparison with M alone.

Levels of hsCRP over time were significantly different (P = 0.0261) among treatment groups (Fig. 1E); M+R dropped from baseline to month12 (P < 0.0001) and remained lower than M and M+L, but values rose from month 12 to 36 in all treatment groups (M, P = 0.0006; M+R, P = 0.0266; M+L, P = 0.0309). Mean homocysteine levels were higher in the M+R group compared with the M group (P = 0.0002) or M+L group (P = 0.0002) (Supplementary Table 1H), but rose significantly from baseline to month 36 (P < 0.0001; Table 1). For PAI-1 and NEFA, no treatment group differences were detected (Supplementary Table 1I and J).

For LDL, non-HDL, and apoB, there was no effect of sex or race-ethnicity across time or treatment group (Supplementary Tables 1A, E, F, 2A, and 2F). For triglycerides, females had significantly lower values than males overall (P = 0.0351), and NHB had lower than H or NHW (both P < 0.0001; Supplementary Table 1B). The percent of participants at high risk (triglycerides ≥150 mg/dL or on fibrate) was also different by race-ethnicity across treatment groups (interaction P = 0.0368; Supplementary Table 2B); in all three treatment arms, high-risk levels in NHB were significantly less prevalent than in both H and NHW, while among NHB, percent at high risk was significantly higher in M than in M+L (P = 0.0133).

Mean HDL levels in females were well below their high-risk cutoff of <50 mg/dL (the majority were at high risk), while mean levels in males were quite close to their cutoff of <40 mg/dL (Table 1 and Supplementary Table 1C and D). The percent in the higher-risk category for sex significantly decreased over time; in females, it fell from 87.6% at baseline to 80.3% at month 36, while males decreased from 65.6 to 57.7% (female vs. male, P < 0.0001; Table 2 and Supplementary Table 2C and D). Mean HDL values in males were significantly different across time by race-ethnicity (interaction, P = 0.0269); by month 36, values were 37.1 and 40.7 mg/dL for H and NHB, respectively (P = 0.0194; neither different from NHW, 40.2 mg/dL).

There were significant differences by sex within treatment group in LDL particle density (interaction, P = 0.0455). In M+R, the percentage with small dense LDL was significantly lower in females than in males (P = 0.0003); in females but not males, percentages in M were higher than in M+R (P < 0.0001) and M+L (P = 0.0150). The sexes were also different across time (interaction, P = 0.0378); females dropped from baseline to month 12 and then remained at the lower level, but males did not vary over time (Supplementary Table 2G). Across treatment arms, NHB had significantly lower percents with small dense LDL than H (P < 0.0001) and NHW (P < 0.0004), even at baseline (NHB, 46.0%; H, 64.5%; and NHW, 63.1%).

For hsCRP, by month 36, levels in females had risen significantly from baseline (0.43 to 0.56 mg/dL; P = 0.0059) compared with males (0.37 to 0.40 mg/dL; not significant). Percentage of females in the high-risk category increased from baseline to month 36 (41.8 to 53.7%; P = 0.0005) and fell in males (39.9 to 33.7%; not significant). NHW had lower mean levels than H (P = 0.0171) or NHB (P = 0.0059), but there were no differences across racial-ethnic groups in percent at high risk. Males had significantly higher homocysteine values than females (P < 0.0001; Supplementary Table 1H). H had higher mean PAI-1 values than NHB (P < 0.0001) or NHW (P = 0.0480) (Supplementary Table 1I). For NEFA, levels in females were significantly higher than those of males (P = 0.0114) and in NHW were significantly higher than in NHB (P = 0.0034) and H (P = 0.0150; Supplementary Table 1J).

Fig. 2 shows regression lines comparing the relationship among HbA_1c and LDL, triglycerides, and HDL (separately by sex) by treatment group. LDL levels rose with increasing HbA_1c levels (P < 0.0001), with no difference across treatment groups (Fig. 2A). The relationship between HbA_1c and triglyceride differed by treatment group (P = 0.0250; Fig. 2B); both M and M+R had significant positive slopes (P = 0.0240 and P = 0.0105, respectively), but M+L was flat, indicating that higher HbA_1c levels were not associated with higher triglyceride levels in this treatment group. HDL was examined separately by sex (Fig. 2C and D). In both females and males, differences by treatment group seen in the figures were not detected statistically (P ≈ 0.08), perhaps due to diminished sample size, although the negative slope for M+R was significant (females, P = 0.0314; males, P = 0.0419), indicating that HDL decreased as HbA_1c increased in this treatment group. BMI was a significant covariate term in all models but did not affect the significance of the relationship between lipid levels and HbA_1c.

At baseline, the TODAY cohort had a remarkably high prevalence of dyslipidemia compared with age-matched nonobese youth and adolescents without diabetes (14,15). More importantly, over the average follow-up of 3.86 years, LDL, non-HDL, and apoB levels, which changed in parallel, triglycerides, and homocysteine levels all rose over time, and a majority of participants had high-risk (low) HDL levels. The percent of TODAY subjects requiring lipid-lowering drugs according to protocol tripled to 10.7%, with a minority achieving LDL goal. In the TODAY study, the only CVD risk factors that appeared to decline over time were the percentages with high-risk HDL and with small, dense, more atherogenic LDL, largely in the first year of follow-up.

Overall, none of the three diabetes interventions prevented the worsening of CVD risk factors over time; however, treatment differences were observed specifically in the proportion of small dense LDL particles and in hsCRP and homocysteine levels, all of which were more atherogenic in the M group. In general, the benefits of a lifestyle intervention, similar in design to that used in TODAY, on lipoprotein and other CVD risk factors in adult patients with prediabetes (16,17) and type 2 diabetes (18) were not seen in the TODAY cohort. The exception was that lifestyle intervention appeared to attenuate the relationship between HbA_1c and triglyceride levels in the entire cohort and between HbA_1c and HDL levels in females. Whether this effect was related to a greater decrease, albeit small and transient, in adiposity (lower BMI, percent fat mass, waist circumference, and abdominal height) in M+L is unknown. The adverse effects of rosiglitazone on conventional lipid levels demonstrated in adults (19) were not seen in TODAY.

Ethnicity and sex had an impact on atherogenic profiles. In all three treatment arms, triglyceride levels in NHB were significantly lower than those in both H and NHW. Among NHB, the percentage with high-risk triglycerides was significantly higher in M than in M+L. These results are consistent with ethnic differences in triglycerides in the general population and suggest triglyceride thresholds regarding metabolic risk may need to be ethnicity-specific (20). Mean HDL levels in females were well below their high-risk cutoff of <50 mg/dL, while levels in males were quite close to their cutoff of 40 mg/dL. Males had significantly higher homocysteine values than females. The addition of either rosiglitazone or lifestyle increased LDL particle size in females.

Limitations of the TODAY study included the variable follow-up over time, although the majority of the cohort had assessments at 36 months. Only three racial-ethnic groups were large enough to provide reliable estimates of their effects. Moreover, the adherence to dietary and exercise management and subsequent statin therapy was likely less in our cohort than is generally seen in adults. Specific adherence data for statin therapy in TODAY are not available. It is possible that worsening lipid profiles were secondary to worsening compliance over time rather than to duration of diabetes. However, while adherence to glycemic control medications waned over time in the TODAY trial, adherence remained >70%; therefore, it is likely that adherence to lipid therapy also remained reasonable in this setting. In addition, hsCRP was higher at baseline in the M group, making it possible that the significantly higher levels of hsCRP after treatment were secondary to randomization effects.

The deteriorating atherogenic and inflammatory risk profile observed in the TODAY cohort over time, despite intensive intervention in the setting of a randomized clinical trial, suggests CVD will become prevalent in the third and fourth decades of life in adolescents with type 2 diabetes. Furthermore, the overall poor psychological health of affected individuals already reported in this cohort will likely further challenge the ability to deliver effective interventions (21). Trials of more aggressive interventions, both pharmacologic and behavioral, to lower CVD risk are necessary in this vulnerable population.

The members of the writing group are as follows: Ruth S. Weinstock (co-chair), MD PhD, SUNY Upstate Medical University; Sonia Caprio (co-chair), MD, Yale University School of Medicine; Kenneth C. Copeland, MD, University of Oklahoma Health Sciences Center; Samuel S. Gidding, MD, Nemours Cardiac Center; Kathryn Hirst, PhD, George Washington University; Lorraine L. Katz, MD, Children's Hospital of Philadelphia; Santica Marcovina, PhD, Northwest Lipid Research Laboratories; Kristen J. Nadeau, MD, University of Colorado Denver; and David M. Nathan, MD, Massachusetts General Hospital.

This work was completed with funding from National Institute of Diabetes and Digestive and Kidney Diseases/National Institutes of Health grants U01-DK-61212, U01-DK-61230, U01-DK-61239, U01-DK-61242, and U01-DK-61254; National Center for Research Resources General Clinical Research Centers Program grants M01-RR00036 (Washington University School of Medicine), M01-RR00043-45 (Children’s Hospital Los Angeles), M01-RR00069 (University of Colorado Denver), M01-RR00084 (Children’s Hospital of Pittsburgh), M01-RR01066 (Massachusetts General Hospital), M01-RR00125 (Yale University), and M01-RR14467 (University of Oklahoma Health Sciences Center); and National Center for Research Resources Clinical and Translational Science Awards grants UL1-RR024134 (Children’s Hospital of Philadelphia), UL1-RR024139 (Yale University), UL1-RR024153 (Children’s Hospital of Pittsburgh), UL1-RR024989 (Case Western Reserve University), UL1-RR024992 (Washington University in St. Louis), UL1-RR025758 (Massachusetts General Hospital), and UL1-RR025780 (University of Colorado Denver).

R.S.W. is site principal investigator with no salary support for industry-sponsored multicenter clinical trials funded by grants administered through the State University of New York Upstate Medical University from Eli Lilly and Company, GlaxoSmithKline, Medtronic, and AstraZeneca. K.C.C. receives a consulting fee and honorarium as a member of the Steering Committee for a research study conducted by Daiichi Sankyo. S.S.G. has a clinical trial of fish oil for high triglycerides in adolescents funded by GlaxoSmithKline. No other potential conflicts of interest relevant to this article were reported.

R.S.W., S.C., K.C.C., K.H., L.L.K., K.J.N., and D.M.N. researched data, contributed to the discussion, wrote the manuscript, and reviewed and edited the manuscript. S.S.G. contributed to the discussion, wrote the manuscript, and reviewed and edited the manuscript. S.M. contributed to the discussion and reviewed and edited the manuscript. K.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Parts of this study were presented at an invited symposium at the 72nd Scientific Sessions of the American Diabetes Association, Philadelphia, Pennsylvania, 8–12 June 2012.

The TODAY Study Group thanks the following companies for donations in support of the study’s efforts: Becton, Dickinson and Company; Bristol-Myers Squibb; Eli Lilly and Company; GlaxoSmithKline; LifeScan; Pfizer; and Sanofi. The TODAY Study Group also thanks the American Indian partners associated with the clinical center located at the University of Oklahoma Health Sciences Center, including members of the Absentee Shawnee Tribe, Cherokee Nation, Chickasaw Nation, Choctaw Nation of Oklahoma, and Oklahoma City Area Indian Health Service for participation and guidance.

Materials developed and used for the TODAY standard diabetes education program and the intensive lifestyle intervention program are available to the public at https://today.bsc.gwu.edu/.