Autonomic Nerve Testing Predicts the Development of Complications: A 12-year follow-up study

From 1990 to 1993, autonomic nerve testing was performed in 335 adolescents with type 1 diabetes. Initial assessment and follow-up of this cohort was approved by the ethics committee at The Children’s Hospital at Westmead. Outcomes examined at follow-up were renal disease, retinal disease, and hypoglycemic awareness. Patients were offered diabetes complications assessment at The Children’s Hospital at Westmead between 2003 and 2005. The first contact was by postal mail, and initial nonresponders were then contacted by phone. If patients declined to participate, or were unable to attend, permission was requested to obtain microvascular complication screening results from other treating doctors. Contact was made with 198 (59%) of the original group; of these, 137 consented to participate in the study. Recruitment and follow-up numbers are shown in Fig. 1.

At the beginning of the study, cardiovascular autonomic nerve function was assessed by three tests of heart rate variation (maximum-minimum heart rate during deep breathing, Valsalva, and lying-standing change) and one of blood pressure (change from lying to standing) (adapted from Ewing et al.) (1). Reference ranges for cardiovascular tests were obtained from 122 nondiabetic control subjects (9). Pupillary autonomic function at baseline and reference ranges were assessed using an infrared computerized pupillometer, as previously described (7,9). The fifth percentile reference range limits used to define abnormal cardiovascular and pupillary tests are described in Table 1 (9).

At follow-up, hypoglycemic awareness was assessed by a validated hypoglycemia unawareness questionnaire (n = 123) (10). Hypoglycemia awareness was scored from 0 to 10 (≤2, aware; 3–6, partially aware; and ≥7, unaware).

Retinopathy was assessed using stereoscopic fundal photography of seven fields in 102 patients. The stereophotographs were taken with a Topcon Fundus Camera (TRC 50-VT; Tokyo Optical, Tokyo, Japan) after dilatation of the pupils with 1% cyclopentolate and 2.5% phenylephrine. Nonsimultaneous photographic pairs were taken of seven standardized fields in each eye and then viewed with a Donaldson Stereoviewer, providing a three-dimensional representation of the fundus, enabling microaneurysms to be more easily distinguished from hemorrhages and artifacts. The Early Treatment Diabetic Retinopathy Study adaptation of the modified Airlie House classification of diabetic retinopathy was used. Mild retinopathy was defined as ≤31, moderate retinopathy was 41–51, and severe retinopathy was ≥61 or any patients who had required laser therapy. The presence or absence of retinopathy as an outcome variable was assessed in the regression analysis. For a further 35 patients, the presence of retinopathy was determined based on external doctors’ reports.

Urinary albumin-to-creatinine ratio was measured in 99 patients. Albumin was measured by competitive chemiluminescence immunoassay by the IMMULITE analyzer (Diagnostic Products, Los Angeles, CA). Creatinine was measured by Jaffe reaction, Dimension ARX (Dade Behring, Newark, DE). Albumin-to-creatinine ratio (in milligrams per millimole) was calculated by the formula [albumin (mg/l)/creatinine (μmol/l)] × 1,000. Microalbuminuria was defined as an albumin-to-creatinine ratio >2.5 mg/mmol.

Results are reported as median (interquartile range). Statistical analysis was performed using SAS (version 8; SAS Institute, Cary, NC). Wilcoxon’s rank-sum tests were used to compare continuous variables between two groups.

Logistic regression was used to assess the association between the outcome (complications at follow-up) and potential explanatory variables. The predictors considered were cardiovascular and pupillometry tests (both as categorical [quartiles] and continuous variables), age, diabetes duration, A1C, blood pressure percentile, and BMI SD scores as continuous variables. Significant variables are reported as odds ratios (ORs) and 95% CIs.

Patient recruitment is shown in Fig. 1. Of those families who were contacted, there were two deaths (one road traffic accident [patient was not driving] and one unknown cause). Seventy-three patients (37%) reported active follow-up with an adult endocrinologist.

At baseline assessment, the median age was 14.7 years (95% CI 13.0–16.8), duration of diabetes 6.3 years (4.0–9.6), and A1C 8.3% (7.5–9.3); at follow-up, the respective values were 26 years (24.7–28.7), 18.9 years (16.1–21.7), and 7.9% (7.3–8.9). Baseline age, duration of diabetes, and complications were similar between the 137 participants (those who participated in the follow-up phase) and 196 nonparticipants (see Table 1). However, baseline A1C was lower (8.2 vs. 8.5%, respectively, P = 0.031).

The median follow-up time was 12.5 years (95% CI 11.8–13.0). No patient had overt nephropathy, was on renal dialysis, or had received a renal transplant. Nineteen patients (19%) had elevated albumin-to-creatinine ratios, which ranged from 2.58 to 23.75 mg/mmol.

Of the 102 patients who had retinal screening results at follow-up, 10 (10%) had severe retinopathy requiring laser therapy, 45 (44%) had mild retinopathy, 15 (15%) had moderate retinopathy, and 32 (31%) had no retinopathy. Of the 70 patients with retinopathy, 19 had retinopathy detected on external doctors’ reports. Although the prevalence of retinopathy was higher in those screened at our institution (73%), versus the external doctors’ reports (59%), the proportions did not differ significantly (P = 0.17).

Hypoglycemia unawareness was present in 7% of patients, while 38% were partially aware and 55% completely aware of all hypoglycemic episodes. In univariate analysis, patients with resting pupil diameter in the lower quartile at baseline were more likely to develop microalbuminuria (OR 2.89 [95% CI 1.02–8.15], P = 0.045) and retinopathy (4.43 [1.22–16.11], P = 0.024) at follow-up (Table 2).

In a multiple logistic regression model after adjusting for baseline A1C and duration of diabetes, resting pupil diameter in the lower quartile (<4.7 mm) predicted the development of both microalbuminuria (OR 4.36 [95% CI 1.32–14.42], P = 0.016) and retinopathy (4.83 [1.3–17.98], P = 0.019) (Fig. 2). A pupillometry test in the lower quartile at baseline was not associated with baseline retinopathy or microalbuminuria.

Baseline cardiovascular tests in the lower quartile were not associated with retinopathy or microalbuminuria 12 years later (17 vs. 21% for baseline vs. follow-up, respectively, P = 0.63). Neither cardiovascular nor pupillometry tests (abnormal or lower quartile) were associated with the development of hypoglycemia unawareness at follow-up.

This longitudinal study identifies a predictive relationship between small pupil size, a marker of early autonomic neuropathy, and the presence of microalbuminuria and retinopathy 12 years later. This relationship persisted after adjusting for glycemic control. Such an association was not identified for conventional Ewing’s test of cardiovascular autonomic function.

We previously reported (6,7) that pupillometry was more sensitive in detecting subclinical autonomic neuropathy than cardiovascular tests in adolescents with type 1 diabetes. More recently, these pupillary tests have been compared in a similar group of young adults (median age 22.8 years and duration of diabetes 16 years) (13). While all pupillary variables differed, none of the conventional Ewing’s battery of cardiovascular tests differed from age-matched control subjects, which further justifies the choice of autonomic nerve tests in our current study. Some of the mathematically derived linear time domain parameters and nonlinear parameters of heart rate variability did differ between patients and control subjects (13).

Another Australian study (8) of adolescents also found an association between pupillometry abnormalities (resting pupil diameter) and microalbuminuria but not cardiovascular tests. The lack of an association between subclinical cardiovascular test dysfunction (in contrast to pupillary dysfunction) and the future development of microvascular complications could be explained by the differential sensitivities of pupillometry and cardiovascular tests in the assessment of autonomic function. The lack of an association between baseline pupillometry and baseline retinopathy or microalbuminuria suggests that pupillometry abnormalities predate detectable retinopathy and microalbuminuria.

It was surprising to find no association between subclinical autonomic neuropathy at baseline and future self-reported hypoglycemia unawareness. The questionnaire that we used had been validated and was found to be a reliable estimation of hypoglycemia unawareness in adults (mean age 38 years) with type 1 diabetes (10). Those with reduced awareness on the retrospective self-reported questionnaire had significantly fewer symptoms and were less likely to detect blood glucose levels <3.9 mmol/l when prospectively recording blood glucose symptoms/estimations on personal hand-held computers (10). It is possible that the younger age of our study group (26 years), in combination with social/cultural differences, may have prevented the accurate reporting of hypoglycemia in our group. In addition, as only seven patients reported hypoglycemia unawareness, our sample size may have been too small to detect an association.

The pathological mechanism linking subclinical neuropathy and future microvascular complications has not been fully clarified. Early glycemic damage to the autonomic nerves may, in parallel, directly damage the retinal and glomerular blood vessels. However, as pupillary dysfunction predicted microvascular complications even after adjusting for glycemic control, we postulate that early autonomic nerve dysfunction affects vascular tone and results in changes to both glomerular and retinal blood flow, thus predisposing to microalbuminuria and retinopathy. Other researchers have hypothesized that damage to the autonomic nerves supplying the renal vasculature may lead to increased renal blood flow, glomerular hyperfiltration, and, thus, microalbuminuria (14). Similarly, increased retinal blood flow during acute hyperglycemia and an overall decreased retinal blood flow in type 1 diabetes may contribute to the development of retinopathy (15). Neurally mediated changes in retinal blood flow may further contribute to the development of retinopathy.

The exact neurological mechanism of damage to the pupillary light reflex (mediated by both sympathetic and parasympathetic nerve fibers) in patients with type 1 diabetes is unclear. The balance between parasympathetic and sympathetic nerve function determines pupil size. Small pupillary diameter may be due to decreased sympathetic or increased parasympathetic activity. Pharmacological testing using cocaine drops have demonstrated sympathetic pathway damage. However, another study (16) reported no difference in the response to cocaine between diabetic and nondiabetic control subjects, whereas pilocarpine caused greater constriction in diabetic patients, suggesting parasympathetic denervation hypersensitivity. In support of initial damage being sympathetic, Hoeldtke et al. (17) have found that sudomotor responses due to acetylcholine are redistributed, i.e., increased in the forearm and decreased in the foot, in diabetic subjects with poor glycemic control.

As baseline A1C was lower in participants than nonparticipants (8.2 vs. 8.5%), we acknowledge that there may be a bias in the follow-up group. However, our clinical experience suggested that the follow-up group attended for a variety of reasons (to obtain a second opinion about existing complications or for an opportunity for reengagement with health services) and were not exclusively the “best” patients. A study of the transition process in the U.K. showed that clinic attendance for young adults with type 1 diabetes fell from 98% pre-transfer to 61% after 2 years (18). Therefore, although it is concerning, it is not surprising that only 37% of our contacted cohort (∼8 years after transition) had active follow-up with an endocrinologist. It is also possible that some of the patients without retinopathy (on external doctors’ reports) did have retinopathy, as we cannot be sure of the quality assurance and reproducibility of retinal photography from external sources. External retinopathy screening methods may not be as sensitive as our own retinal screening, given that 73% had retinopathy when examined by us compared with 59% in the externally examined group. However, the difference in proportions was not statistically significant.

While the relationship between cardiac autonomic neuropathy and microalbuminuria has been described in cross-sectional studies of children with type 1 diabetes (19) and adults with type 2 diabetes (14), our longitudinal study using pupillometry also extends these findings to retinopathy. Our results support the idea that the changes in vascular tone cause damage, in addition to coexisting glycemic damage. With the benefit of an adolescent cohort, we were able to study autonomic neuropathy at an earlier stage of the disease process. Although pupillometry assessment was not performed in the DCCT (Diabetes Control and Complications Trial), it is clear from the DCCT that neuropathy progression can be prevented by improving glycemic control. Therefore, the effect of improved glycemic control on early pupillometry abnormalities needs to be ascertained. In the meantime, it would be prudent to consider these patients at higher risk of developing microalbuminuria and retinopathy.