Treatment of CDO—and indeed all lipid disorders—has historically focused heavily on lifestyle change. Weight loss, ^,,,, changes in dietary composition, ^,,,,,, and changes in physical activity ^,,,,,,,,,,, have all been shown to improve CDO in adults and children in short term studies. Elevated TG levels in adults and children ^,,,,, respond to restriction of carbohydrates. Low HDL-C and unfavorable HDL particle size profiles improve with a fiber-rich nutrient-dense supplement ^, as well as with interventions that increase activity alone or in combination with weight loss. ^,,,,,, Lifestyle change interventions in adults and children not only reduce TG and improve the TG/HDL-C ratio, they also decrease non-HDL-C and lead to LDL particles that are larger and, therefore, more readily cleared. ^{,,,,,,,,,,,,,,,,,} Although these results are encouraging, so far, the impact of these efforts is not large, maintenance is challenging, and lifestyle modification interventions have not been sufficient to address the long-term cardiovascular risk of CDO. Additional therapeutic options are needed.

Their efficacy in children and adolescents with overweight and obesity is unknown. In adults with CDO, statin therapy has been shown to improve both standard lipid and LDL particle profiles by ∼40% to 50%. ^,,,,,,, Statins lower TG and raise HDL, as demonstrated in children ^, and adults, and may be synergistic with lifestyle management. The anti-inflammatory effect of statins may also be important in adiposopathy. The safety and LDL-C lowering effects of statins have been shown in randomized trials of children with familial hypercholesterolemia, and in high-risk pediatric populations including type 1 diabetes mellitus, dyslipidemia in renal insufficiency, renal transplantation, ^, heart transplantation, and systemic lupus erythematosus. However, statins are associated with myopathy and rarely myositis in adults, and concerns have been raised about liver toxicity, although not supported in the literature. Further, trials in adults show statins accelerate the risk for incident type 2 diabetes mellitus in high-risk individuals; ^, one administrative claims database analysis suggests a correlation in childhood. No evidence exists regarding the safety and efficacy of pharmacologic treatment for CDO in childhood, for any statin.

The Dyslipidemia of Obesity Intervention in Teens (DO IT!) Trial is the first study evaluating the use of statin to treat CDO in children and adolescents. This trial sought to determine the efficacy in slowing atherosclerosis and improving lipid parameters, and the safety of statins as an approach to primary prevention of adult cardiovascular disease. In this manuscript, we describe the original study design and methods of the DO IT! trial, including quality control and safety protocols, planned statistical analyses, and recruitment and retention challenges exacerbated by the COVID-19 pandemic. We explain protocol modifications made to try to address these challenges, and share lessons learned, with an eye towards future research study design and implementation.

The primary aim of DO IT! was to compare the effect of pitavastatin calcium 4 mg per day vs placebo, identical in size, shape, and packaging, on vascular measures in adolescents with excess adiposity and CDO. CDO was defined as high non-HDL- C + high TG/HDL-C ratio or low HDL-C and excess adiposity was originally defined as a BMI ≥95th%ile but was adjusted downward over the course of the study, recognizing the primacy of the pathophysiology of adiposity over a broader range of BMI percentiles. Secondary aims compared the effect of pitavastatin vs placebo on safety and lipid outcomes. The study investigators hypothesized that youth treated for 2 years with pitavastatin in addition to standard of care lifestyle advice would have a significant change in vascular measures, reduction of sLDL particles, increase in HDL particles, and no clinically significant adverse safety outcomes when compared with the placebo group.

Pitavastatin calcium, described as pitavastatin in this manuscript, was chosen due to published experience as an LDL-C lowering agent in children with severe dyslipidemia in 1 small ( n = 14), and a second larger pediatric trial (PASCAL Study, n = 106). This larger 12-week double-blinded trial in 10 centers from 6 European countries studied 1, 2, and 4 mg versus placebo ( n = 106), with an open-label extension during which most were up titrated to 4 mg daily ( n = 113). Pitavastatin 4 mg daily resulted in a 39.3% LDL-C reduction, with similar reductions in non-HDL-C and apoB, and no change in HDL-C or TG in this patient population with lipid profiles most typical of familial hypercholesterolemia (eg, normal TG and HDL levels). There were no significant increases in aspartate aminotransferase (AST), alanine aminotransferase (ALT) and creatine kinase (CK), and no differences in adverse events.

Vascular measures carotid femoral pulse wave velocity (cfPWV) and carotid intima-media thickness (cIMT) were chosen as primary outcomes for DO IT! because of their relationship with CDO in adults ^,, and in children, ^,, association in adulthood with childhood CDO, ^,,, progression in PWV in the absence of intervention, studies in adults and in children with familial hypercholesterolemia showing improvement with statins, ^, and because cfPWV and cIMT abnormalities in adulthood predict CVD events. The primary outcome chosen was cfPWV because it is more easily obtained than cIMT, more reproducible, easier to standardize across sites, and responds more quickly to statins. ^,,,,

Secondary outcomes included the standard lipid profile, as the most clinically available test for pediatric practitioners, as well as lipid particle size distribution by nuclear magnetic resonance (NMR) to assess for residual atherosclerosis risk. Safety measures were key secondary outcomes as children with obesity-related dyslipidemia are also at risk for nonalcoholic fatty liver disease (NAFLD), now termed metabolic dysfunction-associated steatotic liver disease, (MASLD), and type 2 diabetes, making it all the more important to assess for signs of adverse physiologic effects.

Patient records were reviewed to determine suitability for screening. At the study opening, patients were considered potentially eligible for screening if they met all the following inclusion criteria based on medical record review or from a recent clinic visit: age 10-17 years, BMI ≥95th percentile by CDC criteria, and a fasting lipid profile (FLP) within the previous 3 months that met study criteria. Exclusion criteria included factors that precluded cooperation with study measures and intervention (eg, developmental deficits), contraindications to statins (eg, evidence of myopathy, liver disease other than MASLD, pregnancy etc..), existing diagnoses that were study safety endpoints (eg diabetes), and factors that impacted vascular measures (eg, hypertension), or were already indications for statins (eg, high LDL suggestive of familial hypercholesterolemia). As is described further below, eligibility criteria evolved over the course of the study. Specifically, the BMI criteria was modified to BMI ≥85th percentile to reflect the growing understanding that the pathophysiology of obesity is not fully represented by BMI ≥95th percentile. Table 1 lists inclusion and exclusion criteria from the protocol at study launch (version 4.1, 1/18/18) and updated criteria in use at study completion (version 7.0, 8/1/20).

The intended sample size for the study was 177 participants randomized per group. Sample size was calculated based on the hypothesis that there would be a significant group (drug vs placebo) by time interaction effect on mean cfPWV in a repeated measures analysis controlled for subject and clinical site-specific random effects. Sample size was computed under several scenarios, assuming baseline cfPWV 6.3 ± 1.1 m/sec, a linear change over time, a correlation within subjects for repeated measures over time of 0.5, a 2-sided alpha of 0.05, 85% power, and 20% dropout. An estimate of 11% decrease in cfPWV with statin treatment and 5% decrease in cfPWV with placebo at 2 years based prior literature ^,, resulted in the need to enroll 177 participants per group for a total target of 354. Alternatively, if the cfPWV remained unchanged in the control group over the 2-year duration, then a sample size of only 100 per group would be required to detect a cfPWV decrease of 8% in the statin group, whereas a sample size of 253 per group would be required to detect a cfPWV decrease of 5% in the statin group.

Prior to study start, an audit of clinical populations available at the original 13 study sites was performed to assess feasibility of study recruitment resulting in an estimated 661 patients meeting eligibility criteria. A consent rate of ∼60% was estimated to achieve the required sample size of 354 randomized participants, thought to be achievable based on consent rates in other pediatric dyslipidemia drug trials. The sample size estimation was inflated to allow for 20% drop-out. Should recruitment prove to be insufficient, a plan was made to add additional study sites.

Informed assent and consent to participate were obtained at the screening visit, which was scheduled to take place no more than 3 months from the prescreening FLP. All inclusion and exclusion criteria were reassessed at the screening visit to confirm eligibility for randomization ( Table 1 ).

Pulse wave velocity (PWV): Arterial stiffness measured by cfPWV was the primary study outcome. cfPWV is the difference in the carotid-to-distal path length divided by the difference in R-wave-to-waveform foot times; higher cfPWV indicates stiffer conduit vessels. A vascular core lab (led by E. Urbina) provided training, certification, quality control and standardized readings of data obtained from the sties. cfPWV measurements were obtained with standard methodology using a SphygmoCor CPV (AtCor Medical, Sydney, Australia); the average of 3 was recorded. ^,,, Repeatability coefficients for PWV were 2.34 m/sec (for mean value of 8.15±3.01 m/sec) with between-observer values of 2.50 m/sec, excellent agreement compared to that seen in healthy children. ^,

Carotid ultrasonography: A second vascular outcome for the trial was carotid intima-media thickness (cIMT). The vascular core lab provided standardized training, certification, quality control and readings. High-resolution B-mode carotid ultrasound was obtained with a linear array vascular 5.0 to 11.0 MHz probe at prespecified angles using a Meyer’s arc. The angle at which it was obtained was noted for reproducibility of measurements. Right and left carotid arteries in the distal 1.0 cm of the common carotid proximal to the bifurcation, the bifurcation itself (“bulb”), and the proximal 1.0 cm of the internal carotid artery were imaged digitally for off-line analyses using an automatic edge detection software. M-Mode measurements of the common carotid were also performed 1 cm proximal to beginning of the carotid bulb. The maximum far wall cIMT of the 3 carotid artery segments were measured on both sides and the mean right and left measurements were used in analyses. ^, M-mode was used to capture the maximal and minimal lumen diameters for calculations of carotid stiffness including Young’s elastic modulus, Peterson’s elastic modulus, ^, beta stiffness index. Analyses of blind duplicate recordings from the vascular core lab demonstrated excellent reproducibility with coefficient of variability (CV) for automatic edge detection for all carotid segments of <4% compared to the older reading techniques such as manual trace (CV 5%) and point-to-point (CV 6%-7%) measures (unpublished data 2014).

The following measures were obtained in the fasting state for all study participants according to the study visit and measurement schedule as described in Table 2 .

Fasting lipid profile (FLP): A total cholesterol (TC), triglycerides (TG), and high-density lipoprotein cholesterol (HDL-C) level was obtained at all study visits and measured by a central core laboratory (University of Michigan) using the RX DAYTONA PLUS CH 8312 enzymatic assay, valid in samples with TG up to 1000 mg/dL. Non-HDL-C was calculated from TC and HDL. Per protocol, TC and LDL-C results were not made available to clinicians as the results could lead to unmasking.

NMR spectroscopy lipoprotein particle assessment: Lipoprotein particle concentrations and size were measured on batched frozen plasma specimens (−70 °C) using proton NMR spectroscopy at a centralized core laboratory (LabCorp Inc., Burlington, NC) at baseline, 6, 12, and 24 months after randomization. Particle concentrations of lipoprotein subclasses were directly obtained from the measured amplitudes of their spectroscopically distinct lipid methyl group NMR signals. ¹⁴⁴ VLDL, LDL, and HDL particle subclasses were quantified from the amplitudes of their spectroscopically distinct lipid methyl group NMR signals using standard methods. ¹⁴⁴

Apolipoproteins AI, B, and CIII (apoAI, apoB, and apoCIII): Apolipoproteins were collected at baseline and 6, 12 and 24 months after randomization and determined by a central core laboratory (University of Michigan) using enzymatic assays.

High-sensitivity C-reactive protein (hs-CRP): Hs-CRP was collected at baseline, 6, 12, and 24 months after randomization and determined by a central core laboratory (University of Michigan) using enzymatic assay.

Participants were assessed for potential adverse events at all study visits, including the screening and baseline visits ( Table 2 ).

Muscle toxicity: CK was measured by each center’s clinical laboratory or certified local lab. Because CK levels vary by clinical laboratory, as well as ethnic origin, age and sex (higher in younger African American men), a safety cut-point of >10 times the upper limit of normal for a study site’s clinical laboratory was used, following the approach of prior trials. Participants were also asked about muscle-related symptoms at each study visit and were instructed to report any symptoms that occurred between visits. If symptoms or CK elevation were noted, an assessment was made for a clinical explanation (eg, muscle trauma, injury, extreme strenuous exercise), and consideration was given to temporarily withholding study drug followed by reassessment, and possibly a retrial of study drug.

Liver toxicity: AST and ALT were measured by each center’s clinical laboratory or at a certified local lab to assess for potential statin effects on the liver and to monitor for development or progression of MASLD. ^, As long the MASLD was not severe (eg, ALT ≥200 u/L), it was not an exclusion criterion for participation. Participants without known liver disease that experienced increases in ALT were assessed and managed as per prespecified algorithm (Appendix), which included an evaluation by the site’s gastroenterologist/hepatologist, as necessary; the results of any additional evaluation were recorded and reported.

Diabetes and glucose homeostasis: Fasting glucose, insulin and HbA1c were measured by each center’s clinical laboratory or at a certified local lab to assess for potential statin effects on incident diabetes and to monitor for impaired glucose homeostasis. An HbA1c ≥6.5% or a fasting glucose ≥126 mg/dL was further evaluated as directed by each center’s endocrinologist; the results of this evaluation and any therapy were recorded. Additionally, insulin and C-peptide were measured by a central core laboratory (University of Michigan) using enzymatic assays at baseline, 6, 12, 18, and 24 months after randomization. Surrogate markers of insulin sensitivity (1/Fasting insulin, HOMA-IR, QUICKI) were calculated. ^,

Basic anthropometric measures were collected including height, weight, BMI, arm and waist circumference, presence of acanthosis nigricans, vital signs, and blood pressure as described in the study schedule ( Table 2 ). Sexual maturity was assessed using Tanner staging questionnaire by validated self-assessment utilizing realistic color photographs of gender-specific Tanner stages. ^,, Dietary assessment and lifestyle behaviors were assessed by completion of the Preventive Cardiology Lifestyle Screener (Appendix) and the Pittsburgh Sleep Quality Index. ^, Each participant was provided with a physical activity tracker device (FitBit), from which 7 days (5 weekdays, 2 weekend days) of data was obtained and included sleep, daily steps, and daily number of activity minutes, averaged over the 7 days of device wear.

Indications for temporary discontinuation of study drug included significant elevations in ALT and other liver abnormalities, as specified in the Appendix, or concern for muscle toxicity including CK elevation >10 times the upper limits of normal or symptoms (muscle pain or weakness) without documented clinical explanation (eg, muscle trauma/injury/excessive exercise) that persisted, or recurred with reintroduction of study drug. Other reasons for study drug discontinuation included incident diabetes defined by the American Diabetes Association criteria (fasting glucose ≥126 mg/dL, HgbA1c ≥ 6.5%, random glucose ≥200 mg/dL, or 2-hr oral glucose tolerance test OGTT glucose ≥200 mg/dL); study drug could be restarted at the discretion of an endocrinologist after appropriate evaluation. Study drug was also discontinued if a participant needed to take a prohibited medication (protease inhibitor, warfarin, colchicine, erythromycin, rifampin, cyclosporine). Finally, study drug could be discontinued if a participant had other adverse events thought to be related to the study drug in the judgment of the study investigator in consultation with the Data Coordinating Center (DCC), PHN Medical Monitor and Data Safety and Monitoring Board (DSMB). Indications for permanent discontinuation of study drug included bariatric surgery, pregnancy or planning pregnancy. Female participants were counseled about pregnancy and appropriate methods of contraception at every study visit. Study participants that developed LDL- C ≥ 160 mg/dL for at least 6 months were, by study protocol, permanently taken off study drug to be treated with a statin (pitavastatin 4 mg per day preferred) at the discretion of the treating physician. In contrast, participants who developed TG ≥500 mg/dL during the trial on at least 2 consecutive assessments were continued on study drug but could be additionally treated at the discretion of their health care provider with a nonstatin lipid lowering agent; if a prohibited medication was chosen, study drug would be permanently discontinued. No unmasking was planned until the end of the study unless requested by the DSMB or in case of an emergency and participants were followed with expected tests and measurements through the planned completion date.

The primary planned analytic approach for all trial outcomes was intention-to-treat (ITT). Each of the vascular outcomes were to be compared between treatment arms at baseline to confirm that randomization had the desired effect and that there were no meaningful differences in these outcomes prior to treatment. Change in the vascular variables would be compared between groups across all study timepoints using hierarchical linear mixed models (PROC MIXED, SAS for Windows® (SAS Institute, Cary, NC)). Planned statistical models included random effects for subject and slope. A test of the treatment group by time interaction term was planned to determine whether the null hypothesis of no difference between treatment-specific slopes over time would be rejected. Further models were to be adjusted for factors known to be associated with the outcome. A sensitivity analysis would be performed using the above model with an additional random effect for clinical site. Similarly, we planned to compare change in each of the lipid variables between groups across all study timepoints using hierarchical linear mixed models including random effects for subject and slope.

Study drug compliance was calculated as (number of tablets taken)/(number of days on study drug). The number of tablets taken was the number of tablets dispensed minus the number of tablets returned. Days on study drug was defined as the date of last bottle log (or date of permanent discontinuation, if earlier) minus date of the first bottle dispensed minus number of days of temporary discontinuation. The number of days prescribed study drug and percent compliance was compared between treatment groups via Wilcoxon test. Reasons for temporary and permanent drug discontinuations were collected.

The DO IT! Trial co-chairs (BM, SdeF) together with PHN leadership were responsible for all aspects of this study. The original DO IT! protocol was reviewed and approved by the PHN Executive Committee and an independent protocol review committee, and all subsequent protocol versions were reviewed and approved by the PHN Data Safety and Monitoring Board, the FDA, Health Canada, and local IRBs of each participating center. Carelon Research served as the PHN Data Coordinating Center (DCC). Site monitoring was performed both in-person and remotely, and a medical monitor reviewed all safety events according to prespecified timelines.

Trial data were collected using a 21 Code of Federal Regulations Part 11-compliant cloud-based electronic data capture (EDC) system with controlled access by roles, audit trails, electronic edits/queries, and electronic signatures (Zelta by Merative). Medical Dictionary for Regulatory Activities (MedDRA) was used for coding of adverse events. Data collected from patient records and study visits were entered directly into the EDC system and routinely monitored via both automatic and manual quality checks for accuracy, logic, and completeness according to standard protocols. After primary analysis, study datasets were further de-identified for the creation of public-use and investigator datasets.

Initial study sites included the 9 core PHN sites (comprising 10 enrolling centers), as well as 3 planned auxiliary sites for a total of 13 centers, 12 in the US and 1 in Canada. Site supervision and engagement followed standard PHN procedures was accomplished through monthly or semimonthly conference calls, semiannual PHN meetings face-to-face (virtual during the COVID19 pandemic), newsletters and emails. Study chairs created outreach resources for site investigators to share at local provider education sessions in cardiology, weight management, and endocrinology clinics. Trial committee meetings provided opportunities to discuss recruitment and retention strategies, protocol modifications, and enrollment. Monthly coordinator meetings were held to field questions about protocol modifications, good clinical practices, regulatory compliance, and various operational issues. Sites with delayed activation received individual attention from study leadership to enhance site progress.