American Journal of Epidemiology
Copyright O 1999 by The Johns Hopkins University School of Hygiene and Public Health
All rights reserved
Vol. 150, No. 11
Printed in USA.
Risk Factors for Progression of Distal Symmetric Polyneuropathy
in Type 1 Diabetes Mellitus
William G. Christen,1 JoAnn E. Manson, 1 ^ Vadim Bubes,1 and Robert J. Glynn, M
for the Sorbinil Retinopathy Trial Research Group5
In a prospective cohort study, the authors examined risk factors for progression of distal symmetric
polyneuropathy (DSP) in type 1 (insulin-dependent) diabetes mellitus. The study population consisted of
participants in the Sorbinil Retinopathy Trial, a randomized trial of aldose reductase inhibition among patients
aged 18-56 years with type 1 diabetes mellitus of 1-15 years' duration. Diagnosis of DSP was based on
standardized clinical neurologic evaluation. A total of 407 participants who did not have definite DSP at
randomization and had at least one follow-up visit were included in the analysis. Stepwise Cox proportional
hazards models were used to examine the independent contribution of baseline variables to progression of DSP.
During follow-up (median, 40 months), 68 participants (17%) showed progression to definite DSP. After
adjustment for age and treatment assignment, independent predictors of progression to definite DSP were total
glycosylated hemoglobin (relative risk (RR) for increase of one percentage point = 1.25; 95% confidence interval
(Cl) 1.12,1.39), height (RR associated with being one inch (2.54 cm) taller = 1.15; 95% Cl 1.05,1.26), cigarette
smoking (ever vs. never) (RR = 1.87; 95% Cl 1.09, 3.21), and female gender (RR = 2.26; 95% Cl 1.09, 4.67).
These data indicate that, in addition to the previously established role for total glycosylated hemoglobin, other
factors including height, cigarette smoking, and female gender may also be independent risk factors for
progression of DSP in type 1 diabetes mellitus. Am JEp/ctem/o/1999;150:1142-51.
diabetes mellitus, insulin-dependent; distal symmetric polyneuropathy; prospective studies; risk factors
As with other long-term complications of type 1 diabetes mellitus, chronic hyperglycemia is believed to
play a role in the etiology of DSP. A number of studies
have shown that elevated glucose levels are associated
with various abnormalities of nerve function, including deficits in conduction velocity (5-11) and vibratory sensation threshold (7, 9, 12, 13). Further, randomized trial data from the Diabetes Control and
Complications Trial (DCCT) clearly indicate that strict
glucose control can delay or prevent the development
and progression of clinically overt neuropathy (14).
The benefit of tight glucose control, however, is not
without notable costs; low blood glucose levels in
response to intensive therapy are associated with a
substantially increased risk of hypoglycemic episodes
and weight gain (15). Data on other potentially avoidable causes of DSP onset and progression, such as high
blood pressure, elevated cholesterol levels, and cigarette smoking, remain sparse.
The conduct of the Sorbinil Retinopathy Trial (SRT)
afforded an opportunity to examine prospectively the
determinants of DSP onset and progression. The SRT
was a multicenter, randomized, placebo-controlled
trial designed to examine the effect of Sorbinil, an
aldose reductase inhibitor, on the course of diabetic
retinopathy, nephropathy, and neuropathy in persons
Distal symmetric polyneuropathy (DSP) is a common, chronic complication of type 1 diabetes mellitus
affecting an estimated 60 percent of type 1 diabetes
patients 30 years of age and older (1). Its pathogenesis,
however, remains unclear and several mechanisms
have been hypothesized (2-4). The identification of
risk factors for DSP may provide insight into the
pathogenesis of this disease and can help to reduce the
morbidity associated with type 1 diabetes mellitus
complications.
Received for publication July 17, 1998, and accepted for publication March 17, 1999.
Abbreviations: Cl, confidence interval; DCCT, Diabetes Control
and Complications Trial; DSP, distal symmetric polyneuropathy;
EDC, Epidemiology of Diabetes Complications Study; RR, relative
risk; SRT, Sorbinil Retinopathy Trial; TGH, total glycosylated hemoglobin (%).
1
Division of Preventive Medicine, Department of Medicine,
Brigham and Women's Hospital and Harvard Medical School,
Boston, MA.
2
Channing Laboratory, Department of Medicine, Brigham and
Women's Hospital and Harvard Medical School, Boston, MA.
3
Department of Epidemiology, Harvard School of Public Health,
Boston, MA.
4
Department of Biostatjstics, Harvard School of Public Health,
Boston, MA.
5
A complete listing of the Sorbinil Retinopathy Trial Research
Group Is available in Arch Ophthalmol 1990; 108:1234-44.
Reprint requests to Dr. William G. Christen, 900 Commonwealth
Ave East, Boston, MA 02215-1204.
1142
Risk Factors for Distal Symmetric Polyneuropathy Progression
with type 1 diabetes mellitus. The primary results concerning the effects of Sorbinil on neuropathy have
been reported previously (16).
MATERIALS AND METHODS
The SRT was a multi-center trial (11 clinical centers), jointly sponsored by the National Eye Institute
and Pfizer Lie, designed and initiated in 1983 to assess
the effect of Sorbinil, an aldose reductase inhibitor, on
the course of diabetic retinopathy, neuropathy, and
nephropathy.
Study population
A detailed description of the methods and organization of this trial appeared in a previous publication
(17). Briefly, participants in the trial were 497 type 1
diabetes mellitus patients aged 18-56 years at study
entry, with duration of insulin treatment between 1 and
15 years, insulin therapy initiated before 41 years of
age, and no major change in insulin treatment regimen
(defined as an increase in the frequency of insulin
injections, or initiation of home blood glucose monitoring or any other major change in management
aimed at improving blood glucose control) during the
3 months before screening. Total glycosylated hemoglobin value >9.25 percent (normal range 6.0-8.8
percent) was required at entry and retinopathy had to
be absent or very mild, with five or fewer microaneurysms and no other abnormalities in each eye (53
percent of participants had retinopathy in at least one
eye at baseline). Female patients were required to be
postmenopausal, to be surgically sterile, or to have an
intrauterine device in place. In addition, patients with
any of the following were excluded: eye disorders
interfering with the evaluation of the retina, severe
debilitating disease, drug or alcohol dependence, creatinine clearance <40 ml/min, hepatic disease or elevated liver function tests, severe atopic disease or
allergies to hydantoins, conditions potentially interfering with the absorption of the drug, Sorbinil, or use of
experimental drugs within the past month.
Neurologic evaluation
A clinical neurologic evaluation was conducted at a
screening visit, and was repeated at the randomization
visit (approximately 3 months after screening) and at
follow-up visits scheduled at 12, 21, 30, 39, and 48
months after randomization. Additionally, during the
final 4 months of the trial (March-June 1988) all participants were scheduled for a final visit. These evaluations were conducted by a study neurologist who was
blinded to treatment group but not necessarily to other
Am J Epidemiol Vol. 150, No. 11, 1999
1143
patient data including lab results or previous exam
results. However, at the time at which the study was
conducted there was little evidence that any of these
factors were related to the risk of neuropathy progression; thus, knowledge of this information was unlikely
to have any material effect on the neurologic evaluation. DSP was determined to be absent, possible, or
definite based on the neurologist's evaluation of the
neurologic history and physical examination. A
determination of definite DSP had to be supported
by at least two of the following findings: symptoms,
abnormal sensory examination, or a pattern of decreased deep tendon reflexes consistent with DSP.
Classification as possible DSP required presence of
one of these findings. Incidence rates of "progression
to definite DSP" (no to definite, or possible to definite)
and of "any progression" (no to possible, no to definite, possible to definite) were calculated.
Patients were excluded from the neuropathy portion
of the trial if any of the conditions listed below were
present at the screening neurologic examination.
1) Diagnosis of definite DSP determined from medical history or from clinical neurologic examination. Such a diagnosis had to be based on at least
two of the following three conditions: symptoms,
abnormal sensory examination, or decreased
deep tendon reflexes, as described above.
2) Neuropathy not secondary to diabetes (e.g., alcohol or drug dependence).
3) Diabetic mononeuropathy or radiculopathy
affecting extremities.
4) Median or ulnar nerve entrapment.
5) Symptomatic peripheral vascular disease (e.g.,
intermittent claudication, ischemic ulcers) that
could interfere with neuropathy assessments.
A total of 56 patients were determined by a study
neurologist to have one or more of the above listed
conditions at the screening visit and were excluded
from the neuropathy portion of the trial. Of the remaining 441 patients, 14 patients were diagnosed with definite DSP at the randomization visit and were excluded
from the present analysis. Also excluded were 20
patients with no follow-up visits after randomization.
Of the remaining 407 patients, possible polyneuropathy was judged to be present in 77 (18.9 percent) at
randomization.
Baseline information
Baseline data included information on smoking status (current, past, never), alcohol intake, blood pressure (diastolic, <70, 70-79, and >80 mmHg; systolic,
<110, 110-119, 120-129, and £130 mmHg), and total
1144
Christen et al.
glycosylated hemoglobin (%) (TGH), expressed as the
mean of two values taken at screening and randomization. Fasting lipid levels were also obtained and
included total cholesterol, low density lipoprotein
(LDL) cholesterol, and high density lipoprotein (HDL)
cholesterol (mg/dl). Levels of total cholesterol were
obtained by enzymatic reaction (18), and LDL and
HDL cholesterol levels were obtained by a precipitation method using magnesium phosphotungstate (19).
Height and weight were measured at randomization.
were selected and included with age, sex, drug treatment, TGH, diabetes duration, systolic blood pressure,
cigarette smoking, and alcohol intake as candidate
variables for regression analyses. We also fit models in
which hypertension (defined as systolic blood pressure
S>140 mmHg, diastolic blood pressure >90 mmHg, or
history of treatment for high blood pressure) was substituted for systolic blood pressure as a candidate variable. For each relative risk, a 95 percent confidence
interval was calculated (20). All p values reported are
two-tailed.
Data analysis
The primary endpoint in this study was progression
to definite DSP defined as the first occurrence during
follow-up of a determination of definite DSP based on
neurologic examination. Analyses were also conducted
for the endpoint of any progression defined as progression of at least one step along the scale of no to
possible to definite DSP.
Statistical analyses used SAS statistical procedures
(20). Life table methods (21) were used to calculate
rates of progression of DSP (progression to definite
DSP and any progression of DSP) in the total sample
for up to 54 months of follow-up. Age-adjusted relative
rates of progression of DSP were calculated for categories of each baseline variable. Incidence rates were
defined as the number of new cases of progression
(progression to definite DSP; any progression of DSP)
divided by the number of person-years of follow-up
within each level of the baseline variable. An individual contributed person-time to total follow-up beginning at baseline and continuing until die occurrence of
definite DSP (or first occurrence of DSP progression in
analysis of the secondary endpoint) or the final followup visit.
Stepwise Cox proportional hazards regression models were used to identify the most significant predictors of progression of DSP (22). Backwards elimination procedures were used and the significance level
for variable removal from the model was set at 0.20
(23). All variables other than gender and cigarette
smoking were evaluated as continuous variables in
regression analyses. A two-level term for cigarette
smoking (ever vs. never) was used to increase statistical power. Terms for age and drug treatment (Sorbinil
or placebo) were included in all models. In instances
where variables were highly correlated (total cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides; height and body mass index), one variable was
selected (based on results from previous studies) for
inclusion as a candidate variable in regression analyses
in order to limit the number of possible predictors, and
thereby avoid overfitting and unreliable risk estimates
(24, 25). Thus, HDL cholesterol (1) and height (26-30)
RESULTS
Baseline characteristics for the 407 SRT participants
included in diis analysis are presented in table 1. Mean
age was 31.4 years (standard deviation (SD) 7.4
years), mean duration of diabetes was 6.5 years (SD
3.5 years), and mean TGH was 11.8 percent (SD 2.0
percent). Seventy-five percent of the participants were
male.
During follow-up (median, 40 mondis), a total of 68
participants (17 percent) showed progression to definite DSP, while 149 participants (37 percent) showed
any progression of DSP from baseline level.
Corresponding life table estimates of the cumulative
hazard of DSP progression over the period of followup were 0.26 for progression to definite DSP and 0.51
for any progression of DSP (table 2).
Age-adjusted relative rates of progression of DSP
according to baseline variables are presented in table
3. Significantly increased risks (p < 0.05) of progression to definite DSP were observed for greater age,
increasing level of TGH, and cigarette smoking. For
any progression of DSP, greater age and increasing
level of TGH were the only significant predictors at
the p < 0.05 level. Elevated levels of diastolic blood
pressure and systolic blood pressure were associated
with increased rates of both DSP endpoints, but none
of these trends reached statistical significance.
The significant independent predictors of DSP progression, identified in stepwise Cox proportional hazards analysis of data for the 401 participants with complete risk factor information at baseline, are shown in
table 4. Variables found to have an independent, significant effect on progression to definite DSP were
TGH, height, cigarette smoking, and sex. All four of
these variables were significant at the p < 0.05 level.
TGH was the strongest predictor of DSP progression.
An increase of one percentage point in baseline level of
TGH was associated with a 25 percent increase in risk
of progression to definite DSP (relative risk (RR) =
1.25; 95 percent confidence interval (CI) 1.12, 1.39).
When we examined TGH quintiles (based on the distribution of TGH in die total sample at baseline) (figure
Am J Epidemiol
Vol. 150, No. 11, 1999
Risk Factors for Distal Symmetric Polyneuropathy Progression
1145
TABLE 1. Baseline characteristics of participants included In the analysis: trie Sorblnil Retinopathy
Trial, 1983
Characteristic
Age (years), mean (SD*)
Diabetes duration (years), mean (SD)
Total grycosylated hemoglobin (%),
mean (SD)
Diastoiic Wood pressure (mmHg),
mean (SD)
Systolic Wood pressure (mmHg),
mean (SD)
Hypertensionf (%)
Cigarette smoking (%)
Never
Past
Current
Alcohol use* (%)
Nondrinker
Ught
Moderate
Heavy
Body mass index (kg/m2), mean (SD)
Total cholesterol (mg/dl), mean (SD)
LDL* cholesterol (mg/dl), mean (SD)
HDL* cholesterol (mg/dl), mean (SD)
Triglycerides (mg/dl), mean (SD)
Male
(n = 306)
Female
(n= 101)
Total
(n = 407)
30.8 (7.7)
6.4 (3.4)
33.0(6.1)
6.7 (3.8)
31.4(7.4)
6.5 (3.5)
11.7(2.0)
12.1 (2.0)
11.8(2.0)
73.4 (8.3)
70.6 (8.5)
72.7 (8.4)
120.0(12.2)
3.0
111.1 (10.9)
9.2
117.8(12.5)
7.6
54.6
21.9
23.5
49.5
26.7
23.8
53.3
23.1
23.6
44.1
36.3
15.0
4.6
24.0 (3.4)
188.3(41.2)
119.0(33.3)
47.0(13.6)
110.8(129.4)
62.4
29.7
6.9
1.0
24.1 (3.4)
186.2(34.9)
113.8(32.3)
56.3(14.4)
76.8 (37.8)
48.6
34.6
13.0
3.7
24.0 (3.4)
187.8(39.7)
117.7(33.1)
49.3 (14.4)
102.4(114.7)
* SD, standard deviation; LDL, low density lipoprotein; HDL, high density lipoprotein.
t Defined as systolic Wood pressure £140 mmHg, diastoiic blood pressure 290 mmHg, or history of treatment
for high blood pressure.
t Light drinker, 1-9 g/day; moderate drinker, 10-29 g/day; heavy drinker 230 g/day.
TABLE 2. Lrfe-table estimates of progression of distal symmetric potyneuropathy (DSP) In the Sorblnll Retinopathy Trial
Month
Progression to
definite DSP*
Any
progression!
12
21
30
39
48
54
0.01
0.05
0.10
0.16
0.26
0.26
0.06
0.19
0.27
0.35
0.46
0.51
* No to definite, possible to definite.
t No to possible, no to definite, possible to definite.
1), patients in the highest quintile (TGH > 13.5) had a
sixfold increased risk of progression to definite DSP
compared with patients in the lowest quintile (TGH <
10.2) (trend test, p < 0.0001) after adjustment for age,
treatment group, and other predictors retained in the
final Cox model (height, smoking, sex).
Height was also an important independent predictor
of progression to definite DSP. As indicated in the final
stepwise Cox model, an increase of one inch (2.54 cm)
in height was associated with a 15 percent increase in
the risk of progression to definite DSP (RR = 1.15, 95
percent CI 1.05, 1.26). When we considered categories
of height, relative risks of progression to definite DSP
Am J Epidemiol
Vol. 150, No. 11, 1999
for increasing categories of height (inches (cm))
inches (<167.6 cm) (referent), 67-69 inches
(170.2-175.3 cm), 70-72 inches (177.8-182.9 cm),
>72 inches (>182.9 cm) were 1.0 (referent), 1.44, 1.83,
2.46, respectively (trend test, p < 0.01) (data not
shown). The relative risks for increasing categories of
height, defined separately for males and females, are
displayed in figure 2. For each sex, patients in the
highest category for height compared with those in the
lowest category had an approximate two- to threefold
increase in risk of progression to definite DSP. The test
of trend across categories of height was statistically
significant in males (p = 0.01) but not in females (p —
0.09), perhaps due to the small number of cases among
females (21 cases of progression to definite DSP).
Cigarette smoking and female gender were also
identified as independent risk factors for progression
to definite DSP. Patients who were ever smokers, compared with never smokers, had almost twice the risk of
progression to definite DSP (RR = 1.87; 95 percent CI
1.09, 3.21). Females had an approximate twofold
increase in risk of DSP progression compared to males
(RR = 2.26; 95 percent CI 1.09, 4.67).
When we considered any progression of DSP (no to
possible, no to definite, possible to definite), only TGH
and systolic blood pressure were retained in the final
1146
Christen et al.
TABLE 3. Age-adjusted relative risks (RR) and 95% confidence Intervals (Cl) for progression of distal symmetric potyneuropathy
(DSP) according to baseline variables: trie Sorblnll Retinopathy Trial, 1983
Any progression
Progression to definite DSP
Baseline variable
Age (years)
<30
30-39
40-^9
£50
Sex
Female
Male
Total gtycosyiated hemoglobin
<10.2
10.2-11.1
11.2-12.1
12.2-13.5
>13.5
Diastolic blood pressure
(mmHg)
<70
70-79
£80
Systolic blood pressure
(mmHg)
<110
110-119
120-129
£130
Hypertension*
No
Yes
Body mass Index (kg/m2)
<22
22-23
24-26
£27
No.
No.
RR
95% a
(trend)
No.
RR
95% Cl
175
180
48
4
19
33
14
2
1.0
1.8
4.0
10.6
(referent)
1.0,3.1
2.0, 8.1
2.4, 45.9
0.0001
54
66
26
3
1.0
1.2
2.3
3.9
(referent)
0.9, 1.8
1.4, 3.7
1.2,12.5
0.0005
101
306
21
47
1.0
0.9
(referent)
0.5, 1.4
0.55
44
105
1.0
0.8
(referent)
0.6, 1.2
0.27
92
83
86
73
71
5
13
15
13
22
1.0
2.6
2.8
2.7
6.4
(referent)
0.9, 7.4
1.0,7.6
1.0, 7.7
2.4, 16.8
21
33
28
33
34
1.0
1.8
1.4
2.0
2.8
(referent)
1.1,3.2
0.8, 2.6
1.1,3.5
1.6,4.8
0.0007
124
199
83
16
36
16
1.0
1.5
1.9
(referent)
0.8, 2.7
0.9, 3.7
36
80
1.0
1.5
0.07
32
1.5
(referent)
1.0,2.2
0.9, 2.4
0.07
102
133
102
69
14
22
20
12
1.0
1.3
1.9
1.5
(referent)
0.6, 2.5
0.9, 3.7
0.7, 3.2
1.0
1.1
1.3
0.15
35
46
38
29
(referent)
0.7,1.6
0.8, 2.0
0.8, 2.1
0.21
375
31
64
4
1.0
0.7
(referent)
0.3, 1.9
0.50
138
10
(referent)
0.4, 1.6
0.53
106
126
107
63
14
24
18
11
1.0
1.2
(referent)
0.6, 2.3
0.5, 2.1
0.5, 2.0
1.0
0.88
31
49
38
29
1.1
1.3
(referent)
0.8, 2.1
0.7,1.8
0.8, 2.1
0.41
118
1.0
119
53
18
20
20
10
1.5
(referent)
0.6, 2.2
0.6, 2.3
0.7, 3.3
0.33
47
41
42
18
1.0
0.9
0.9
0.9
(referent)
0.6,1.4
0.6, 1.4
0.5, 1.6
0.75
235
172
40
28
1.0
0.9
(referent)
0.5,1.4
84
65
1.0
0.51
(referent)
0.7,1.4
0.99
1.0
1.0
0.0001
1.3
1.0
0.8
1.3
(trend)
Height (inchest)
566
67-69
70-72
>72
Diabetes duration (years)
1-6
7-15
114
1.2
1.2
1.0
Table continues
model as significant independent predictors of progression, and only TGH exceeded the conventional
level for statistical significance of p < 0.05. An
increase of one percentage point in baseline TGH level
was associated with a 17 percent increase in risk of any
progression of DSP (RR = 1.17; 95 percent Cl 1.08,
1.26). Relative risks for increasing quintiles of TGH
were 1.0 (referent), 1.8, 1.5, 2.0, and 2.8 (trend test, p
< 0.001, data not shown) For systolic blood pressure, a
10 mmHg increase was associated with a 9 percent
increase in risk of any progression of DSP (RR = 1.09;
95 percent Cl 0.97, 1.24).
We also fit stepwise Cox models in which we evaluated hypertension in place of systolic blood pressure
as a candidate variable. Thirty-one participants were
categorized as hypertensive. For progression to definite DSP, substituting hypertension for systolic blood
pressure resulted in the same final model as that listed
in table 4. When we considered any progression of
DSP, only TGH (RR = 1.16; 95 percent Cl 1.08, 1.25)
was retained in the final model (in addition to the
forced variables of age (RR = 1.05; 95 percent Cl
1.02, 1.07) and drug (RR = 0.89; 95 percent Cl 0.64,
1.23)).
Am J Epidemiol Vol. 150, No. 11, 1999
Risk Factors for Distal Symmetric Polyneuropattiy Progression
TABLE 3.
1147
Continued
Progression to definite DSP
Baseline variable
1*5.
Any progression
n
No.
RR
95% Cl
H
No.
RR
95% Cl
69
37
43
1.0
1.0
1.3
(referent)
0.7, 1.6
0.9, 2.0
0.18
75
50
(referent)
0.7, 1.4
0.6, 1.6
0.4, 2.3
0.85
(trend)
Cigarette smoking
Never
Past
Current
Alcohol usef
Nond linker
Light
Moderate
Heavy
Total cholesterol (mg/dl)
^162
163-183
184-209
>209
HDL§ cholesterol (mg/dl)
<39
39-47
48-56
>56
LDL§ cholesterol (mg/dl)
<36
97-114
115-139
>139
Triglycerides (mg/dl)
<58
59-77
78-110
>110
217
94
96
23
19
26
1.0
1.5
2.2
(referent)
0.8, 2.9
1.2,3.9
198
141
41
19
53
15
6
1.0
0.7
0.6
0.6
(referent)
0.4, 1.2
0.3, 1.4
0.2, 2.7
1.0
1.0
0.9
1.4
(referent)
0.5, 2.2
0.4, 1.9
0.7, 2.8
1.0
1.0
0.7
0.7
(referent)
0.5, 1.8
0.4, 1.5
0.4, 1.5
14
1.0
13
17
24
0.9
1.0
1.3
(referent)
0.4, 1.9
0.5, 2.1
0.7, 2.6
17
11
1.0
0.8
19
21
1.1
1.5
2
109
15
101
13
16
24
104
93
94
105
96
112
102
100
110
92
116
98
101
92
18
19
14
17
(referent)
0.4, 1.7
0.6, 2.2
0.8, 2.9
0.008
(trend)
0.13
5
1.0
1.0
1.0
0.9
0.33
39
34
37
39
1.0
0.9
0.9
1.0
(referent)
0.5, 1.4
0.5,1.4
0.6, 1.6
0.97
1.0
0.29
37
39
33
40
(referent)
0.6,1.5
0.5, 1.4
0.5, 1.3
0.41
19
1.0
0.8
0.9
37
1.0
33
(referent)
0.6, 1.4
0.6, 1.5
0.6, 1.6
0.96
(referent)
0.8, 2.0
0.7,1.8
0.8, 2.1
0.31
0.39
39
40
0.9
1.0
1.0
0.15
38
35
38
38
1.0
1.3
1.1
1.3
• Defined as systolic blood pressure £140 mmHg, dlastolic blood pressure £90 mmHg, or history of treatment for high Wood pressure.
X 1 inch = 2.54 cm.
t Light drinker, 1-9 g/day; moderate drinker, 10-29 g/day; heavy drinker, 230 g/day.
§ HDL, high density lipoprotein; LDL, low density lipoprotein.
DISCUSSION
These prospective data demonstrate a strong association between level of glycemia, as measured by
TGH, and the risks of progression of DSP. Patients
in the highest quintile of TGH were six times more
likely than those in the lowest quintile to show progression to definite DSP. The data also indicate that
patient height, cigarette smoking, and female gender may be important risk factors for the development of DSP. Duration of diabetes, blood pressure
level, and blood lipid levels were not significantly
related to the risks of progression of DSP in this
population.
This study had several important strengths. The SRT
consisted of one of the largest groups of insulindependent diabetics to be examined prospectively. It
employed a high standard of neuropathy detection
through the use of standardized neurologic evaluations
conducted at a screening visit, at randomization, and
during regularly scheduled follow-up visits. The study
Am J Epidemiol
Vol. 150, No. 11, 1999
population was comprised of high risk persons due to
the requirement that participants have a TGH value at
baseline >9.25 g/dl. Duration of diabetes, however,
was relatively short-term (range 1-15 years) in this
population. As a result, only 14 percent of patients had
to be excluded from the present analysis because of a
diagnosis of definite DSP at baseline (or because of
evidence of other conditions precluding involvement
in the neurologic component of the trial).
There are few other prospective studies of patients
with type 1 diabetes mellitus with which to compare
rates of DSP. The incidence of DSP in the SRT appears
to be comparable with that reported in the Pittsburgh
Epidemiology of Diabetes Complications (EDC)
Study (30) (cumulative probability of definite DSP
from life table analysis: SRT, 0.26 at 4.5 years; EDC
Study, 0.29 at 6 years). The SRT rates also appear to be
generally comparable with those reported in the DCCT
(secondary intervention cohort, conventional therapy
group) (14). However, despite the similarity of rates,
1148
Christen et al.
TABLE 4. Predictors of progression of distal symmetric
poryneuropathy (DSP) (final model selected In backwards
stepwlse proportional hazards regression analysis): the
Sorbinil Retinopathy Trial*
Covartate
Relative
risk
95% Clt
P
value
Progression to definite DSP
Age
Drug*
TGHt
Height
Ever smoker
Female gender
1.06
1.05
1.25§
1.15H
1.87
2.26
1.03, 1.10
0.66, 1.70
1.12, 1.39
1.05, 1.26
1.09,3.21
1.09, 4.67
0.0004
0.84
0.0001
0.003
0.023
0.028
Any progression Of DSP
Age
Drug*
TGH
Systolic blood
pressure
1.05
0.87
1.17§
1.02, 1.07
0.63, 1.21
1.08,1.26
0.0001
0.40
0.0001
1.09#
0.97, 1.24
0.15
• Age and drug were forced in all models.
t Cl, confidence interval; TGH, total glycosylated hemoglobin,
i Sorbinil relative to placebo.
§ Relative risk for a one percentage point increase in TGH.
Tl Relative risk associated with increase of one inch in height.
# Relative risk associated with 10 mmHg increase in systolic
blood pressure.
these three cohorts differ in several respects. As in the
SRT, participants in the DCCT (secondary intervention
cohort) were required to have had type 1 diabetes mellitus for 1-15 years (mean duration: SRT, 6.5 years;
DCCT, 8.8 years). However, participants in SRT had
either no retinopathy or very mild non-proliferative
retinopathy (microaneurysms only), whereas in the
secondary intervention cohort of the DCCT, participants were required to have very mild-to-moderate
non-proliferative retinopathy. Glycemic levels were,
on average, slightly higher in the SRT than in the
DCCT (TGH, 11.8 percent vs. 11.1 percent (estimated)), and the SRT population was slightly older
(mean age, 31.4 years vs. 27.0 years). The EDC Study
population (as described in the analysis of risk factors
for DSP (30)) was younger (mean age, 25.1 years) than
either the SRT or DCCT cohort, but had markedly
greater duration of type 1 diabetes mellitus (mean
duration, 16.9 years). Glycemic levels in the EDC
(mean TGH, 10.3 percent) were somewhat lower than
in the SRT or DCCT cohort.
The finding in the SRT of a strong association
between glycemic control and progression of DSP is
consistent with available evidence from other studies.
The DCCT demonstrated that with intensive therapy,
designed to achieve glycemic levels as close to normal
as possible, the development of confirmed clinical
neuropathy can be reduced by two-thirds (14). Our
data indicate a clear dose-response relationship
between level of glycemia and risks of DSP progression suggesting that even modest reductions in level of
glycemia may be of benefit in reducing the risks of
DSP progression. The possibility of graded benefits
Totri flrtmn
(n-401)
P. trend - 0.0001
&0
5.0
4.0
ao
20
1.0
ao
<10.2
10.2-11.1
11.2-12.1
12.2-13.5
Total Glycosytated Htmoglobtn (%)
>13.6
FIGURE 1. Relative risks for progression to definite distal symmetric polyneuropathy (DSP) according to baseline total glycosylated hemoglobin (%) quintiles adjusted for age, drug treatment assignment, height, and cigarette smoking: the Sorbinil Retinopathy Trial, 1983.
Am J Epidemiol Vol. 150, No. 11, 1999
Risk Factors for Distal Symmetric Polyneuropathy Progression
Famalaa
(N-100)
P, trend = 0.09
Males
(N = 301)
P, trend = 0.01
<67
67-69
70-72
1149
>72
Height (Inches)
<63
63-64
>64
FIGURE 2. Relative risks for progression to definite distal symmetric polyneuropathy (DSP) according to categories of height in males and
females adjusted for age, drug treatment assignment, total glycosylated hemoglobin (%), and cigarette smoking: the Sorbinil Retinopathy Trial,
1983.
with glucose control would have important implications in the clinical management of diabetes and its
complications, especially in view of the increased risks
of hypoglycemic episodes associated with intensive
therapy (15). The specific mechanisms mediating the
association of glycemic level and DSP remain unclear.
Peripheral nerve tissue does not require insulin for glucose uptake and cytoplasmic glucose concentrations
generally reflect the degree of glycemic control.
Chronically elevated glucose levels may increase the
risks of DSP through activation of the polyol pathway
(31), by causing hypoxia (32), or perhaps through nonenzymatic glycosylation of structural proteins and the
accumulation of advanced glycosylation end products
(33, 34).
Height was also an important independent predictor
of progression to definite DSP in the SRT. In clinical
studies of persons with diabetes, greater height has
been associated with a higher frequency of abnormalities of nerve function, including deficits in nerve conduction velocity (27, 29) and vibratory perception (26,
27). Our data indicate that height also predicts the
development of neuropathy; men taller than 72 inches
(182.9 cm) were more than twice as likely to develop
DSP compared with men less than 67 inches (170.2
Am J Epidemiol Vol. 150, No. 11, 1999
cm) in height. A similar trend, although not statistically
significant, was observed in women. Other epidemiologic studies also tend to support an association of
height with DSP. Studies include cross-sectional data
from the DCCT (28) and recently reported prospective
data from the EDC Study (30). In the latter study, a one
standard deviation increase in height (3.6 inches (9.1
cm)) was associated with a doubling of the risk of DSP
(in a final stepwise regression model that included type
1 diabetes mellitus duration, TGH, cigarette smoking,
and hypertension). In the SRT, a one standard deviation increase in height (3.7 inches (9.4 cm)) was associated with a somewhat smaller 66 percent increase in
the risk of DSP (in a final model that included age,
treatment group, TGH, cigarette smoking, and sex).
Patients with DSP exhibit axonal loss which increases
in severity distally (36), and DSP is usually evident
first in the legs and feet, perhaps due to the greater
length of the distal axons in the lower limbs. Thus,
height may be a proxy for neuron length (26). Greater
neuron length may increase vulnerability to metabolic
or transport dysfunction (36).
The SRT data also indicated that those who had ever
smoked, compared with never smokers, had almost
twice the risk of DSP. Comparable results were
1150
Christen et al.
reported in the EDC Study in which ever smokers had
a statistically significant 73 percent increased risk of
developing DSP compared with never smokers (30). A
somewhat larger effect for ever smokers (RR = 2.5
among patients with type 1 diabetes mellitus; no association observed among patients with non-insulindependent diabetes mellitus) was reported in a small
case-control study conducted in a referral-based diabetes clinic (37). There are several plausible mechanisms for an effect of cigarette smoking on progression
of DSP. Cigarette smoking increases carboxyhemoglobin (38) and platelet adhesiveness (39-41), and
adversely affects blood lipid levels (42), thereby
increasing the viscosity of the blood and decreasing
arterial flow. Cigarette smoke also causes vasoconstriction of peripheral arteries (43). All of these
changes can lead to tissue hypoxia and, perhaps ultimately, to many of the abnormalities in peripheral
nerve normally associated with diabetes.
Finally, female gender was associated with a twofold
increased risk of developing definite DSP in the SRT.
This finding is compatible with prospective data from
the EDC Study in which females had a slight, but statistically nonsignificant, elevated risk of DSP (30). The
finding is not in agreement, however, with crosssectional data from the DCCT. In that study, patients
with clinically detectable neuropathy at baseline were
significantly more likely to be male (28). Additional
research is required to determine if there are genderrelated differences in the risk of developing DSP. Such
a finding could indicate a modulatory effect of sex hormones in the pathophysiology of DSP, or, alternatively,
might result from gender differences in self-reporting
of symptoms or exposure to non-diabetic risk factors
for neuropathy.
Several variables identified as predictors of DSP in
other populations did not emerge as independent predictors of DSP in the SRT. Duration of diabetes was
not an important determinant of DSP in the SRT, possibly reflecting the limited range of duration in this
cohort. We also found little evidence for an important
role for blood pressure, even though in the EDC Study
hypertension (defined as systolic blood pressure of
>140 mmHg, diastolic blood pressure of >90 mmHg,
or currently taking blood pressure medication)
emerged as the strongest risk factor for DSP. In the
SRT, systolic blood pressure was a weak predictor {p
— 0.15) of DSP when we considered any progression
of DSP from baseline level, but was not predictive of
progression to definite DSP. Nor was hypertension
selected as a predictor for either endpoint when it was
substituted for systolic blood pressure as a candidate
variable. The reasons for these discrepant findings are
not clear. The SRT selection criteria may have pro-
duced few participants with high blood pressure
thereby limiting the power of the study to detect an
effect of this factor on neuropathy. In any event, the
resultant wide confidence intervals for elevated blood
pressure do include fairly large positive effects consistent with previous reports.
In summary, these prospective data from a cohort of
individuals with relatively short-term duration type 1
diabetes mellitus indicate a strong and graded relation
between glycemic control and risk of DSP. The data
also provide further evidence that cigarette smoking is
an important, avoidable cause of DSP progression.
These results are most directly applicable to other type
1 diabetes mellitus patients who have little or no
retinopathy or pre-existing definite DSP, but who are
otherwise at high risk for development and progression
of diabetic complications. Furthermore, the findings
regarding an increased risk with greater height and
with female gender may ultimately be of clinical value
in the identification and targeting for specific treatment of subgroups at high risk for this common complication of type 1 diabetes mellitus.
ACKNOWLEDGMENTS
The Sorbinil Retinopathy Trial was originally funded by
Pfizer, Inc., and the National Eye Institute. This work was
also supported by Determinants of Diabetic Retinopathy in
Sorbinil Retinopathy Trial, National Eye Institute, grant no.
EY10624.
REFERENCES
1. Mascr RE, Steenkiste AE, Dorman JS, et al. The epidemiologic
correlates of diabetic neuropathy: a report from the Pittsburgh
Epidemiology of Diabetes Complications Study. Diabetes
1989;38:1456-61.
2. Clements RS. Diabetic neuropathy—new concepts of its etiology. Diabetes 1979;28:604-ll.
3. Brown MJ, Asbury AK. Diabetic neuropathy. Ann Neurol
1984;15:2-12.
4. Editorial: Diabetic neuropathy. Lancet 1989;1:1113—14.
5. Gregersen G. Diabetic neuropathy: influence of age, sex, metabolic control, and duration of diabetes on motor conduction
velocity. Neurology 1967;17:972-80.
6. Pietri A, Ehle A, Raskin P. Changes in nerve conduction velocity after six weeks of glycoregulation with portable insulin
infusion pumps. Diabetes 1980;29:668-71.
7. Boulton AJM, Drury J, Clarke B, et al. Continuous subcutaneous insulin infusion in the management of painful diabetic
neuropathy. Diabetes Care 1982;5:386-90.
8. Young RJ, Ewing DJ, Clarke BF. Nerve function and metabolic control in teenage diabetics. Diabetes 1983;32:142-7.
9. Service FJ, Rizza RA, Daube JR, et al. Near normoglycaemia
improved nerve conduction and vibration sensation in diabetic
neuropathy. Diabetologia 1985;28:722-7.
Am J Epidemiol Vol. 150, No. 11, 1999
Risk Factors for Distal Symmetric Polyneuropathy Progression
10. Dahl-Jorgensen K, Brinchmann-Hansen O, Hanssen KF, et al.
Effect of near nonnoglycaemia for two years on progression of
early diabetic retinopathy, nephropathy, and neuropathy: the
Oslo study. Br Med J 1986;293:1195-9.
11. Ziegler D, Mayer P, Muhlen H, et al. The natural history of
somatosensory and autonomic nerve dysfunction in relation to
glycaemic control during the first 5 years after diagnosis of
Type 1 (insulin-dependent) diabetes mellitus. Diabetologia
1991 ;34:822-9.
12. Holman RR, Mayon-White VM, Orde-Peckar C, et al.
Prevention of deterioration of renal and sensory-nerve function
by more intensive management of insulin-dependent diabetic
patients. A two-year randomized prospective study. Lancet
1983; 1:204-8.
13. Jakobsen J, Christiansen JS, Kristoffersen I, et al. Autonomic
and somatosensory function after two years of continuous subcutaneous insulin in type I diabetes. Diabetes 1988;37:452-5.
14. The Diabetes Control and Complications Trial Research
Group. The effect of intensive diabetes therapy on the development and progression of neuropathy. Ann Intern Med 1995;
122:561-8.
15. The Diabetes Control and Complications Trial Research
Group. The effect of intensive treatment of diabetes on the
development and progression of long-term complications in
insulin-dependent diabetes mellitus. N Engl J Med 1993;
329:977-86.
16. Sorbinil Retinopathy Trial Research Group. The Sorbinil
Retinopathy Trial: neuropathy results. Neurology 1993;43:
1141-9.
17. Sorbinil Retinopathy Trial Research Group. A randomized trial
of sorbinil, an aldose reductase inhibitor, in diabetic retinopathy. Arch Ophthalmol 1990; 108:1234-^4.
18. Allain CC, Poon LS, Chan CSG, et al. Enzymatic determination of total serum cholesterol. Clin Chem 1974;20:470-5.
19. Lopes-Virella MF, Stone P, Ellis S, et al. Cholesterol determination in high-density lipoproteins separated by three different
methods. Clin Chem 1977;23:882-4.
20. SAS Institute Inc. SAS procedure guide, Release 6.03 edition.
Cary, NC: SAS Institute Inc, 1988.
21. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Statist Assoc 1958;53:457-81.
22. Cox DR. Regression models and life tables. J R Stat Soc Series
B 1972;34:187-220.
23. Maldonado G, Greenland S. Simulation study of confounderselection strategies. Am J Epidemiol 1993;138:923-36.
24. Concato J, Feinstein AR, Holford TR. The risk of determining
risk with multivariable models. Ann Intern Med 1993;118:
201-10.
25. Harrell FE, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and
adequacy, and measuring and reducing errors. Stat Med 1996;
15:361-87.
26. Sosenko JM, Gadia MT, Foumier AM, et al. Body stature as a
risk factor for diabetic sensory neuropathy. Am J Med 1986;
80:1031-4.
27. Gadia MT, Natori N, Ramos LB, et al. Influence of height on
quantitative sensory, nerve conduction, and clinical indices of
Am J Epidemiol
Vol. 150, No. 11, 1999
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
1151
diabetic peripheral neuropathy. Diabetes Care 1987; 10:
613-16.
The DCCT Research Group. Factors in development of diabetic neuropathy. Baseline analysis of neuropathy in feasibilty
phase of Diabetes Control and Complications Trial (DCCT).
Diabetes 1988;37:476-81.
Duck SC, Wei F, Parke J, et al. Role of height and glycosylated
hemoglobin in abnormal nerve conduction in pediatric patients
with type I diabetes mellitus after 4-9 yr of disease. Diabetes
Care 1991;14:386-92.
Forrest KYZ, Maser RE, Pambianco G, et al. Hypertension as
a risk factor for diabetic neuropathy. A prospective study.
Diabetes 1997;46:665-70.
Sima AAF, Nathaniel V, Bril V, et al. Histopathological heterogeneity of neuropathy in insulin-dependent and noninsulin-dependent diabetes, and demonstration of axoglial
dysjunction in human diabetic neuropathy. J Clin Invest 1988;
81:349-64.
Dyck PJ. Hypoxic neuropathy: does hypoxia play a role in diabetic neuropathy? The 1988 Robert Wartenberg Lecture.
Neurology 1989;39:111-18.
Vlassara H, Brownlee M, Cerami A. Nonenzymatic glycosylation of peripheral nerve protein in diabetic mellitus. Proc Natl
AcadSci 1981;78:519O-2.
Brownlee M, Cerami A, Vlassara H. Advanced glycosylation
end products in tissue and the biochemical basis of diabetic
complications. N Engl J Med 1988;318:1315-21.
Thomas PK. Diabetic neuropathy: epidemiology and pathogenesis. In: Pickup JC, Williams G, eds. Chronic complications of diabetes. Oxford: Blackwell Scientific Publications,
1994:101-11.
Arezzo JC, Schaumburg HH, Spencer PS. Structure and function of the somatosensory system: a neurotoxicological perspective. Environ Health Perspect 1982;44:23-30.
Mitchell BD, Hawthorne VM, Vinik Al. Cigarette smoking
and neuropathy in diabetic patients. Diabetes Care 1990;13:
434-7.
Nordenberg D, Yip R, Binkin NJ. The effect of cigarette smoking on hemoglobin levels and anemia screening. JAMA 1990;
264:1556-9.
Ogston D, Bennett NB, Ogston CM. The influence of cigarette smoking on the plasma fibrinogen concentration.
Atherosclerosis 1970; 11:349-52.
Hawkins RL. Smoking, platelets and thrombosis. Nature 1972;
236:450-2.
Levine PH. An acute effect of cigarette smoking on platelet
function. A possible link between smoking and arterial thrombosis. Circulation 1973;48:619-23.
Craig WY, Palomaki GE, Haddow JE. Cigarette smoking and
serum lipid and lipoprotein concentrations: an analysis of published data. Br Med J 1989;298:784-8.
Holbrook JH, Grundy SM, Hennekens CH, et al. Cigarette
smoking and cardiovascular diseases. A statement for health
professionals by a Task Force appointed by the Steering
Committee of the American Heart Association. Circulation
1984;7O:1114-17.