Decline in Neurophysiological Function After 7 Years in an Adolescent Diabetic Cohort and the Role of Aldose Reductase Gene Polymorphisms

Patients were Australian adolescents with type 1 diabetes attending their annual complications screening at the Children’s Hospital at Westmead. Inclusion criteria were two or more neuropathy assessments performed over a minimum of 4 years, with the median time between the first and last assessment of 7.0 years (interquartile range 5.1–11.1). Two hundred and sixty-two patients satisfied these criteria. Genomic DNA was extracted from white blood cells for the genotyping of the AKR1B1 gene. All patients gave informed consent, and the study was approved by the ethics committee of the Children’s Hospital at Westmead, Sydney, Australia.

Cardiovascular autonomic function was assessed by the measurement of heart rate variation during deep breathing (DBHRV), the Valsalva maneuver, and during standing from a lying position (30-to-15 ratio). Pupillometry was performed in the left eye after 5 min of dark adaptation with an infrared computerized pupillometer (Pupilscan; Fairville Medical Optics, Amersham, Buckinghamshire, U.K.). The phasic light reflex was measured in response to a light stimulus of 25-foot candles by recording pupillary diameter over the next 3 s. Resting pupil diameter was measured, and the maximum constriction velocity and reflex amplitude were calculated.

Thermal thresholds for hot and cold were determined on the dorsum of the left foot using a thermal threshold tester (Medelec, Old Working, U.K.), and vibration thresholds were determined at the large toe and medial malleolus of the left foot (Biothesiometer; Biomedical Instrument, Newbury, OH), as previously described, using reference ranges derived for 11- to 19-year-old subjects (18,19).

Genomic DNA was extracted from white blood cells for genotyping of the AKR1B1 gene for the following polymorphisms: AKR1B1 promoters (C-106T and C-12G), (CA)_n dinucleotide repeats (microsatellites), and intragenic BamH1 polymorphism.

The 263-bp fragment from the promoter region, containing two single nucleotide polymorphisms (C-106T and C-12G), and the (CA)_n dinucleotide repeat were amplified for all 262 patients as previously described (20,21). The BamH1 polymorphism was identified by using the restriction enzyme BamH1 (NEB). After digestion, the B-allele (wild type) yielded the undigested 252-bp fragment and the b (A–C substitution)-allele yielded 174- and 78-bp fragments (22).

Genotyping for all polymorphisms was repeated on a second occasion to ensure interexperiment consistency. To ensure intersampling consistency, two blood samples were obtained, on two separate occasions from 13 patients, and genotypes were identical.

Longitudinal analysis was performed by using generalized estimating equations with all available data points included for each patient in SPIDA software (version 6; Statistical Computing Laboratory, Sydney, Australia). All neurological outcome variables (three pupillary, three cardiovascular, and two peripheral nerve) were analyzed continuously, with the effect modification of genotypes/alleles on changes in nerve function evaluated over time. The effect of genotypes/alleles was assessed with adjustment for duration, time, sex, HbA_1c (A1C), and blood pressure percentiles when appropriate. Statistical significance was defined as P < 0.05.

At first assessment of 262 patients (114 male patients, 45%), the median age was 12.7 years (interquartile range 11.7–13.9), median duration was 5.3 years (3.4–8.0), and median A1C was 8.5% (7.8–9.3). Ethnic background was Caucasian (n = 234, 89%), Middle Eastern (10, 4%), Asian (9, 3%), or other. Over the median follow-up of 7 years, the median number of assessments per patient was five (four to seven) and A1C was 8.5% (7.6–9.3). At last assessment, their median age was 20.1 years (17.9–24.0) and median duration of diabetes was 13.5 years (9.9–17.4). Mild to moderate background retinopathy was present in 126 of 257 (49%) and microalbuminuria in 11 of 245 (4.5%) subjects.

All three pupillary function tests and two peripheral nerve tests declined over time. Resting pupil diameter reduced from 6.25 to 5.91 mm (P = 0.003), reflex amplitude reduced from 2.07 to 1.73 mm (P < 0.0005), and maximum constriction velocity decreased from 6.23 to 5.33 mm/s (P < 0.0005). Thermal discrimination declined over time, with the hot threshold increasing from 0.31 to 0.74°C (P < 0.0005) and cold threshold increasing from 0.12 to 0.21°C (P < 0.013). Vibration thresholds did not change over time. Only one cardiovascular autonomic test, DBHRV, declined from 35 to 30 beats/min (P < 0.0005). At last assessment, 44% had abnormal pupillary tests, 14% had abnormal cardiovascular autonomic tests, 30% had abnormal thermal threshold tests, and 7% had abnormal vibration thresholds.

The Z-2 allele of the (CA)_n repeat polymorphic marker was more common in this study population compared with the Z+2 allele (58 vs. 24%, respectively) (Table 1). This is in agreement with published results from other type 1 and type 2 diabetic populations with microvascular complications (11,15,23).

Over time, patients with at least one copy of the Z-2 allele had slower maximum constriction velocity, indicative of worse autonomic nerve function, compared with patients without the Z-2 allele (Fig. 1, P = 0.045). Patients who were homozygous for the Z-2 allele had reduced maximum constriction velocity compared with patients without the Z-2 allele (P = 0.026).

For peripheral sensory nerve tests, the presence of the Z+2 allele was associated with a smaller increase in thermal threshold for heat, indicating better sen-sory nerve function, compared with other microsatellite alleles over time (Fig. 2). In addition, a smaller increase in hot thermal threshold was also found significantly associated with Z+2/Z+2, Z+2/-106T, and bb (BamH1) alleles.

The genotype ZZ/-106CC had a smaller reduction in DBHRV, indicating better autonomic nerve function compared with those without the genotype. Similarly, the Z+2/Z+2 genotype was associated with a smaller increase in cold threshold (Table 1). The promoter polymorphism C-12G was not associated with any difference in nerve function test results.

In this study of 262 adolescents with median diabetes duration of 6 years at baseline, neurophysiological function declined during follow-up. Both autonomic and quantitative sensory nerve function declined over a median of 7 years. Polymorphisms in the AKR1B1 gene were significantly associated with both increased and decreased rates of decline of neurological complications in these patients.

The nerve function decline may in part be due to an age effect. When reference ranges were derived in our unit in adolescents aged 11–19 years, DBHRV was negatively correlated with age, but no such correlation was found for the other variables measured (18,19). This study does show a decline in nerve function in patients with type 1 diabetes in this age-group (12–24 years), using noninvasive tests, specifically DBHRV, the resting pupil size, and the phasic light response and discrimination thresholds for hot and cold. Vibration discrimination did not decline, but the frequency of abnormalities was low, measured by biothesiometry, suggesting that this test is not as sensitive as thermal discrimination in this young age-group.

From our results, it was evident that pupillometry measurements were more likely to deteriorate over time than the cardiovascular autonomic measurements. Of the cardiovascular tests, DBHRV was more sensitive than the other two measurements. This is in agreement with the Diabetes Control and Complications Trial, in which intensive therapy significantly slowed the decline in DBHRV but not the Valsalva maneuver (5). Others (7,24,25) have found pupillary tests, both resting and dynamic tests, to be more sensitive than cardiovascular tests to a diabetes effect. Our data also could be interpreted as evidence that the cardiovascular autonomic nervous system is affected later than the pupil.

The reported quantitative sensory tests are easier to administer and less invasive than nerve conduction velocity studies. Thermal threshold discrimination tests small unmyelinated and thinly myelinated fibers. These tests are independent of the observer/operator but are limited by the subject’s attention, motivation, and cooperation. However, any variation is unlikely to have contributed systematically to the increase in abnormalities over time. Indeed, a learned effect from frequent testing should show improvement and not deterioration of the tests. Furthermore, the operators were blinded to patient’s AKR1B1 genotypes.

Patients with the Z-2 allele of the (CA)_n repeat polymorphism exhibited greater decline in maximum constriction velocity of pupil, i.e., worse autonomic nerve function, compared with patients without the Z-2 allele. Conversely, patients with the Z+2 allele or Z+2 with the −106T genotype showed attenuated decline in thermal discrimination, i.e., better nerve function, compared with patients lacking these alleles. In addition, the bb genotype of the intragenic BamH1 polymorphism also significantly associated with slower decline of hot thermal discrimination.

Most of the observed modifications of neurological responses support Z-2 as being the “susceptible” allele and Z+2, Z+2/T, and bb as being the “protective” allele/genotypes. This is in agreement with previous cross-sectional association studies (9,16,22,23) of microvascular complication studies and, more specifically, in neuropathy. Cross-sectional studies, particularly in patients with shorter diabetes duration, can provide only limited information on the association between complications and genes.

This is the first longitudinal genetic association study in relation to diabetic neuropathy in type 1 diabetic patients reported to date. Functional correlates of these polymorphisms are in agreement with the susceptible and protective polymorphisms at the (CA)_n locus shown in this study. The Z-2 allele of the (CA)_n repeat polymorphism has been reported to be associated with higher aldose reductase mRNA levels (26) and with higher aldose reductase activity levels (27). An in vitro study by Hodgkinson et al. (28) found the susceptible Z-2 allele had higher transcriptional activity compared with the protective Z+2 allele. Subsequently, Yang et al. (29) reported that Z-2 with −106C had higher transcriptional activity compared with Z+2 with the −106T haplotype of the AKR1B1 gene. Only the DBHRV protection association with the ZZ/CC genotype is surprising. This genotype appeared protective in the current study but was previously classified by us and others within the susceptible genotype group (9,29). Further functional studies are needed to clarify its role. The precise localization of the polyol pathway in small fibers is uncertain. Aldose reductase immunoreactivity was detected in the Schwann cell cytoplasm of unmyelinated fibers of peripheral rat nerve (30). However, no information is available on localization of sorbitol dehydrogenase in unmyelinated fibers. In myelinated fibers, there appears to be more sorbitol dehydrogenase in axons than in Schwann cells (14).

Our findings are somewhat at variance to a recent longitudinal study of type 2 diabetic patients. Sivenius et al. (11) found that type 2 diabetic patients with the −106T allele had lower sensory response amplitude measurements compared with −106C at baseline examination. However, associations were no longer significant after allowing for physiological and environmental factors, specifically BMI. The observation at baseline can be related to the significantly higher BMI for the patients with the −106T compared with patient with the −106C allele. Moreover, in a larger group, we have recently showed increased BMI and height to be positively associated with peripheral nerve function abnormalities in adolescents with type 1 diabetes (31).

Familial clustering of diabetes complications has been shown for nephropathy and also retinopathy (32,33) but has not been demonstrated for neuropathy. Duration of diabetes before complication onset is important in detecting gene associations (major and minor susceptibility genes [34]). Our cohort of patients was unselected and asymptomatic and was recruited from routine complication assessment visits. Futhermore, the statistical model enables adjustment for duration and glycemic control and allows for varying length of follow-up and number of assessments.

Determining the best methods for measuring neuropathy progression is essential for the design of appropriate, cost-effective clinical trials. In addition, pharmacogenomics can aid in successful design of trials to identify patients with particular risks. Quantitative sensory testings and pupillometry are valuable tools that could be used along with genotyping in interventional studies.

The authors gratefully thank Pfizer for unrestricted grant support.