Impact of Diabetes and Its Treatment on Cognitive Function Among Adolescents Who Participated in the Diabetes Control and Complications Trial

Between 1983 and 1989, 1,441 subjects with type 1 diabetes were enrolled in the DCCT. A total of 249 subjects were recruited as adolescents aged 13–19 years: 32% were 13–14 years old, 37% were 15–16 years old, and 31% were 17–19 years old. All adolescents had to be at least Tanner stage II in pubertal development, which is the stage at which the first signs of puberty are visible on physical examination. We chose age 19 as the upper age limit, rather than age 18 as used in other studies on the DCCT cohort (15,16), because the sample size was considerably larger when the age limit was extended to the final year of adolescence. The DCCT consisted of two cohorts. The primary prevention cohort (n = 149) had diabetes for 1–5 years, no retinopathy, and urinary albumin excretion <40 mg/24 h. The secondary intervention cohort (n = 100) had diabetes for 1–15 years, very mild to moderate nonproliferative retinopathy, and urinary albumin excretion ≤200 mg/24 h at baseline. Approximately half of the adolescent sample (n = 115) was randomly assigned to intensive therapy (three or more insulin injections daily or subcutaneous infusion with an external pump, guided by frequent self-monitoring of blood glucose) with preprandial blood glucose level targets between 3.9 and 6.7 mmol/l, a monthly A1C target within the nondiabetic range (<6.05%), and a goal of avoiding hypoglycemia. The remainder (n = 134) was assigned to conventional therapy with one to two daily insulin injections and no numeric blood glucose targets but freedom from symptoms of hyperglycemia and from frequent or severe hypoglycemia as the therapeutic goal. At the end of the DCCT, this cohort of patients had been studied for an average of 7.3 years (range 4–10). Intensive therapy was recommended for all subjects because it had been shown to be highly effective in reducing complications of long-term diabetes. Subjects in the conventional treatment group were given training in aspects of intensive therapy and then returned to their own health care providers. Between April 2004 and May 2006, 175 participants (76% of surviving, eligible participants) were reevaluated with the cognitive test battery; 74 participants who were adolescents at the DCCT baseline did not participate in the Epidemiology of Diabetes Interventions and Complications (EDIC) follow-up cognitive testing. Of these, 6 had died and 12 were inactive at the time of testing.

Cognitive testing, as originally described for the DCCT (2), was performed at each site by personnel who were trained and certified by the DCCT/EDIC Central Neuropsychological Coding Unit. The test protocol is described elsewhere (2,4). Standardized tests were administered in a fixed order. Capillary blood glucose levels were routinely monitored immediately before testing and at its midpoint to rule out hypoglycemia during testing. If a subject was found to have a blood glucose level ≤3.89 mmol/l, testing was stopped and the patient was given a snack; after waiting at least 15 min, testing was resumed when the reading returned to at least 5.0 mmol/l. Tests scoring procedures are described elsewhere (2,4).

During the DCCT, 24 test variables were chosen a priori to be of particular diagnostic value when applied to patients with type 1 diabetes, and a standardized (Z) score was calculated for each, with the mean ± SD from the baseline assessment of the DCCT cohort used as a reference (2) to provide a unit-free measurement of the relative improvement or deterioration in performance compared with the total group at baseline. Details of the test variables and domains are described elsewhere (4).

During EDIC, subjects completed an annual history, physical examination, electrocardiogram, and laboratory testing, including serum creatinine and hemoglobin A1C, using the same methods as during the DCCT (17). Participants reported the presence of sensory symptoms of peripheral neuropathy as part of neuropathy screening (18).

Psychiatric symptomatology was assessed with the Symptom Checklist-90-Revised (SCL-90R), which was administered annually during the DCCT and once in the EDIC in the same year that the cognitive testing was performed (19,20). For this report, the depression scale was used to assess the effects of mood on cognition.

During the DCCT, severe hypoglycemia was defined as any event requiring the assistance of another individual, including seizure or coma, with either blood glucose <2.78 mmol/l and/or subsequent reversal of symptoms with oral carbohydrate, glucagon injection, or intravenous glucose (1). For the purposes of this article, severe hypoglycemic events are limited to those leading to coma and/or seizure because these episodes are the most likely to have an impact on cognition and are most precisely defined. At quarterly visits, study coordinators asked about the occurrence of hypoglycemia since the last visit, and all such events were reported to the Data Coordinating Center as soon as possible after their occurrence. During the EDIC, the severe hypoglycemic events that occurred in the 3 months before the annual visit were documented on the annual history form, and further details surrounding these events were recorded.

Demographic and clinical characteristics were compared with the use of Wilcoxon's rank-sum test to evaluate the differences between the treatment groups for ordinal and numeric variables (21). The contingency χ² test was used for categorical variables; when the sample size was small, Fisher's exact test was used (21). All treatment group comparisons were based on intention to treat.

Separate analysis of covariance models were used to assess the effects of treatment group (intensive or conventional), mean A1C values stratified by tertiles (<7.9, 7.9–9.5%, and >9.5%), and frequency of severe hypoglycemia (0, 1–5, and >5 reported events) on the standardized quantitative score for each of the eight cognitive domains. Each model adjusted for baseline age, sex, education, length of follow-up, visual acuity, self-reported sensory loss attributable to peripheral neuropathy, and the number of interval cognitive tests taken (to control for practice effects). Results are presented as the average increase or decrease in the standardized score from the DCCT baseline within or between groups or as the per unit change in a quantitative covariate. Nominally significant results (P < 0.01) are cited.

Analyses were repeated to determine whether cognitive performance was associated with 1) current mood state measured by the SCL-90R depression scores (scores ≥63 are suggestive of a possible depressive disorder), 2) the timing of severe hypoglycemia, and 3) diabetic ketoacidosis (DKA) during the DCCT.

Table 1 presents the characteristics of the patients at the DCCT baseline and at the EDIC year-12 follow-up. There were no statistically significant between-group differences at the DCCT baseline. The characteristics presented in Table 1 were also compared between patients who continued participation in the EDIC cognitive follow-up and those subjects who were still actively participating in other EDIC evaluations but did not participate in the EDIC cognitive evaluation. The only statistically significant difference was severe nonproliferative diabetic retinopathy at EDIC year 12 (17% participants and 42% nonparticipants). Of the patients who did not participate in the EDIC cognitive follow-up, 45% were assigned to intensive treatment. Furthermore, 16 of 38 had severe nonproliferative diabetic retinopathy, and 12 of 41 had peripheral neuropathy at EDIC year 12. No data on these variables were available for the remaining nonparticipants.

At EDIC year 12, the age of participants ranged from 29 to 41 years (mean ± SD 35.2 ± 2.5 years). Of the participants, 40% reported having completed a college degree (37% intensive and 42% conventional), and almost 50% reported a professional or technical occupation (53% intensive and 41% conventional). At EDIC year 12, differences between the two treatment groups approached significance for the presence of peripheral neuropathy and severe nonproliferative diabetic retinopathy (P < 0.05).

Over the entire 18-year follow-up, there were a total of 294 reported episodes of coma or seizure. Of these, 200 were reported in 51 intensive treatment group subjects and 94 were reported in 36 conventional treatment group subjects (Table 2).

A separate analysis conducted between the primary and secondary patient cohorts differing in terms of diabetes duration and complications revealed no effect of diabetes duration between these two groups. Table 3 summarizes the raw scores for each cognitive test, stratified by treatment group. Figure 1 shows cognitive test results for each domain by original treatment group (Fig. 1A), cumulative number of severe hypoglycemic events (0, 1–5, and >5 episodes) (Fig. 1B), and metabolic control (tertiles of mean A1C values) (Fig. 1C). Neither original treatment assignment nor cumulative number of hypoglycemic events influenced performance in any cognitive domain. Age did not significantly affect cognition. Performance in both sexes improved over time, and this effect was more pronounced in male participants (P < 0.01). Higher values of A1C were associated with modest declines in psychomotor and mental efficiency (P < 0.01). Degree of self-reported symptoms of depression at year 12, as indexed by the SCL-90R (median T score 46.0; scores ≥63 reflect possible depressive disorder), was not associated with poorer performance on any of the eight domains. Analyses were repeated using the broader definition of hypoglycemia, which includes episodes in which the patient has incapacity sufficient to require assistance. The results with the broad definition were similar to those obtained using the narrow definition (i.e., restricted to seizure or coma).

The timing of severe hypoglycemic events did not affect performance on any of the eight cognitive domains. We found that 47 patients reported their first episode of coma or seizure during adolescence (ages 13–19) and 43 patients reported having lost consciousness between one and five times before their DCCT baseline evaluation. Further, there was no synergistic effect of hypoglycemia and hyperglycemia on cognition. Patients who experienced DKA (n = 26) during the DCCT (ages 13–19 years) declined in cognitive performance on the learning domain, whereas the patients without DKA improved. Further, patients with DKA episodes during the DCCT improved less on the spatial information domain than patients with no DKA events (data not shown).

Our previous report on the entire DCCT/EDIC cohort showed no detrimental effects of intensive treatment or severe hypoglycemic episodes on cognitive performance (4). However, because of the potential vulnerability of the developing brain (14), we evaluated whether intensive treatment during adolescent years posed threats to long-term cognitive functioning. Severe hypoglycemic episodes during childhood are a major concern, especially given the findings that cognitive deficits may be more common in those in whom type 1 diabetes is diagnosed during childhood (9,12,22). Moreover, children may be more sensitive than adults to even modestly lower glucose levels, with cognitive deterioration at ∼3.3 versus 2.5 mmol/l in adults (23).

Our results closely resemble those previously reported for the entire cohort (4). Intensive treatment is not associated with risk for long-term cognitive dysfunction, even in the subset of patients who entered the DCCT during adolescence. As with the findings from the entire cohort, we found that higher A1C values were associated with poorer performance on measures of psychomotor and mental efficiency, which require the integration of motor and cognitive processes. This finding further highlights the benefits of intensive glycemic control. Higher A1C values were also associated with somewhat slower performance on a simple measure of motor speed (Fig. 1), but, unlike our earlier results with the entire cohort, that effect failed to reach statistical significance in this cohort, possibly due to smaller sample size. Although we collected data on retinopathy, which are associated with higher A1C levels, the effect that this complication has on cognitive ability is beyond the scope of the article and will be addressed separately.

Despite no discernible ill effects on cognition as a result of severe hypoglycemia, brain abnormalities due to serious hypoglycemia have been observed in other studies. For example, children with a history of severe hypoglycemia have shown some abnormalities in brain structure and function (22). Slow-wave electroencephalographic activity is increased in this patient population, especially in the frontal regions (24), which govern executive function and attention. Finally, recent evidence has suggested that severe hypoglycemia in children alters gray matter density (22), analogous to what has been reported in young adults with type 1 diabetes (25). There is not always a direct relationship between brain changes and behavioral effects. Thus, although no notable cognitive deficits were observed, these early brain changes may serve as a marker of future cognitive impairments (25).

Although our conclusion from this study is that severe hypoglycemic episodes have no long-term effect on cognition, even when experienced during adolescence, several study limitations need to be considered. First, we included patients between the ages of 13 and 19 years; therefore, we cannot determine whether diabetes or hypoglycemia during early childhood is associated with cognitive deficits later in life. Accordingly, we have limited information on the effects of diabetes on the very young brain. Second, the results must be generalized cautiously not only because the sample size of our cohort is relatively small but also because of the careful selection criteria applied to subjects recruited into the DCCT. Third, our study imposed restrictions on the number and severity of DKA and hypoglycemic episodes that patients could experience during the several years before the DCCT. These exclusionary factors restrict our ability to comment on whether long-term cognition is further affected in patients who experienced these particularly serious metabolic consequences of diabetes.

In summary, our study indicates that with regard to long-term cognitive function, intensive treatment is safe for patients who had the diagnosis of diabetes as children, despite the increased threat of severe hypoglycemic episodes. Nevertheless, we need to remain cognizant of the dangers of acute hypoglycemia, which can lead to comas, accidents, injuries, death, family stress, loss of school or work time, and loss of commitment to the goals of intensive treatment. Thus, continued research is needed to develop new therapies and technologies that will minimize or eliminate this major obstacle to achievement of optimal control of type 1 diabetes.

Funding for this study was provided by grant 5 R01 DK062218-02 from the National Institute of Diabetes and Digestive and Kidney Diseases and by contracts with the Division of Diabetes, Endocrinology and Metabolic Diseases of the National Institute of Diabetes and Digestive and Kidney Diseases, the General Clinical Research Centers Program, National Center for Research Resources, and the Herbert Graetz Psychosocial Research Fund.