ORIGINAL
ARTICLE
E n d o c r i n e
C a r e
Rate of ␤-Cell Destruction in Type 1 Diabetes
Influences the Development of Diabetic Retinopathy:
Protective Effect of Residual ␤-Cell Function for More
Than 10 Years
Koji Nakanishi and Chizuru Watanabe
Department of General Internal Medicine and Metabolism, Toranomon Hospital, Kawasaki 213-8587, Japan; and the Okinaka Memorial
Institute for Medical Research, Tokyo 105-8470, Japan
Context: Although residual ␤-cell function delays the onset and progression of diabetic retinopathy in patients with type 1 diabetes, the rate of ␤-cell destruction is variable.
Objective: The aim of the study was to clarify the influence of the rate of ␤-cell destruction on the
development and progression of diabetic retinopathy in type 1 diabetes.
Design: We performed a historical cohort study regarding residual ␤-cell function and retinopathy.
Setting: The study was conducted in the outpatient clinic of a general hospital.
Patients: A total of 254 patients with type 1 diabetes participated.
Main Outcome Measures: Serum C-peptide and fundus findings were evaluated longitudinally.
Results: The cumulative incidence of mild nonproliferative diabetic retinopathy was higher in the
patients without detectable ␤-cell function than in those with residual ␤-cell function at 20, 15, and
10 yr after the onset of diabetes (P ⫽ 0.013, P ⫽ 0.006, and P ⫽ 0.048, respectively), but not at 5 yr
after the onset (P ⫽ 0.84). There were higher mean glycosylated hemoglobin values during the
entire follow-up period in the patients without detectable ␤-cell function at 20 and 15 yr after the
onset of diabetes (P ⫽ 0.030 and P ⫽ 0.042, respectively). Positivity for HLA-A24 and -DQA1*03, as
well as the acute onset of diabetes, was associated with early ␤-cell loss and also with early development of diabetic retinopathy. Cox proportional hazards analysis showed that undetectable
␤-cell function at 20, 15, or 10 yr after the onset of diabetes was an independent risk factor for the
development of diabetic retinopathy.
Conclusions: Undetectable ␤-cell function within 10 yr of the onset of type 1 diabetes is associated
with the earlier occurrence of diabetic retinopathy. (J Clin Endocrinol Metab 93: 4759 – 4766, 2008)
S
everal studies have shown that the existence of residual ␤-cell
function delays the onset and progression of diabetic retinopathy in patients with type 1 diabetes through better glycemic
control (1– 4). However, residual ␤-cell function changes over
time in type 1 diabetes, and the rate of ␤-cell destruction is variable both before (5, 6) and after the onset of clinical diabetes (6).
Some patients eventually show complete loss of ␤-cell function,
whereas residual ␤-cell function is maintained in others over a
long period (6). In previous studies that examined the relationship between residual ␤-cell function and diabetic complications,
␤-cell function was assessed in a cross-sectional manner (2) or
over a short follow-up period (3, 4), so the longitudinal changes
of ␤-cell function were not addressed.
Recently, we documented the longitudinal changes of residual ␤-cell function in a cohort of patients with type 1 diabetes (6).
In the present study, we retrospectively investigated the time
0021-972X/08/$15.00/0
Abbreviations: CI, Confidence interval; HbA1c, glycosylated hemoglobin; HLA, human
leukocyte antigen; HR, hazard ratio; NPDR, nonproliferative diabetic retinopathy; PDR,
proliferative diabetic retinopathy.
Printed in U.S.A.
Copyright © 2008 by The Endocrine Society
doi: 10.1210/jc.2008-1209 Received June 4, 2008. Accepted September 19, 2008.
First Published Online September 30, 2008
J Clin Endocrinol Metab, December 2008, 93(12):4759 – 4766
jcem.endojournals.org
4759
4760
Nakanishi and Watanabe
␤-Cell Destruction and Diabetic Retinopathy
course of retinopathy in relation to longitudinal changes of residual ␤-cell function in the same patient cohort and evaluated
how many years of residual ␤-cell function were required to
protect patients with type 1 diabetes from the early onset and
progression of diabetic retinopathy.
Subjects and Methods
Subjects
A total of 254 patients with type 1 diabetes [145 men and 109 women
aged 34 ⫾ 14 yr (mean ⫾ SD) at the onset of diabetes], who presented to
Toranomon Hospital for the first time between establishment of the
hospital in 1957 and the end of 2002 and were periodically followed
thereafter, were evaluated in the present study. A diagnosis of type 1
diabetes was made according to the American Diabetes Association
guidelines (5). In addition, urinary C-peptide excretion of less than 6.6
nmol/d or an integrated serum C-peptide value during the 100-g oral
glucose tolerance test of less than 3.3 nmol/liter was used to define type
1 diabetes, as described previously (2, 6). Autoantibodies to glutamic
acid decarboxylase 65 were positive at the onset of diabetes in 80.2%
(105 of 131) of the patients tested. This study was approved by the
Committee for Investigations Involving Human Subjects of Toranomon
Hospital. All patients gave informed consent for DNA analysis and Cpeptide measurement.
Assessment of the time course of ␤-cell destruction
A sensitive C-peptide RIA was used to assess residual ␤-cell function
(7). A fasting serum C-peptide level below the detection limit (0.017
nmol/liter), a level less than 0.033 nmol/liter at 2–3 h postprandially, or
a serum C-peptide response of less than 0.033 nmol/liter after a 100-g
oral glucose load was defined as loss of detectable ␤-cell function (6, 8).
This serum C-peptide level was originally selected as a cutoff value for
discrimination between complete ␤-cell destruction and the presence of
minimal residual ␤-cell function (6, 8), but a previous cross-sectional
study also showed a difference in the incidence of retinopathy between
patients stratified by this cutoff value (2).
Measurement of serum C-peptide was done in 241 patients. The sera
were stored at ⫺80 C until the assay, which was performed yearly since
our first report on C-peptide assay (7). In 30 patients, the first assessment
of residual ␤-cell function revealed loss of detectable ␤-cell function at
more than 5 yr after the onset of diabetes, so they were excluded from
longitudinal observation of residual ␤-cell function to ensure that the
error in determining the time of loss of detectable ␤-cell function was less
than 5 yr (6). Longitudinal observation of residual ␤-cell function was
performed in the other 211 patients, as described previously (6). The
clinical onset of diabetes was defined as the starting point. In 184 patients, serum C-peptide was measured a total of 5.1 ⫾ 3.2 times (mean ⫾
SD) (range, 2–26 times) over a disease duration of 13.5 ⫾ 10.6 yr (mean ⫾
SD) (range, 0.08 - 49 yr). In 19 patients, loss of detectable ␤-cell function
was found by their first assessment of ␤-cell function within 5 yr after the
onset of diabetes. In eight patients, the presence of residual ␤-cell function was only investigated by one test at 4 –24 yr (median, 14 yr) after the
onset of diabetes, so data for these patients were censored at that time.
The calendar year at the onset of diabetes was earlier in the patients
excluded from the longitudinal study than those included in the longitudinal study [1975 ⫾ 6 vs. 1984 ⫾ 11 (mean ⫾ SD); P ⬍ 0.0001 by
Mann-Whitney U test].
Evaluation of diabetic retinopathy
Evaluation of diabetic retinopathy was performed in 236 patients as
described previously (2). To evaluate yearly the presence or progression
of diabetic retinopathy, patients with type 1 diabetes have been referred
to an ophthalmologist in Toranomon Hospital, to whom C-peptide status was unknown. Optic fundi were examined using indirect ophthal-
J Clin Endocrinol Metab, December 2008, 93(12):4759 – 4766
moscope after papillary dilatation with 0.5% tropic amide. Seven ophthalmologists participated in fundus examination through the study
period. They sketched the findings in a medical chart in the consistent
manner in which the author (K.N.) was instructed. Grading of diabetic
retinopathy was performed by the review of medical records by the author (K.N.). The grading of retinopathy in our previous study (2, 9, 10)
was adapted to that in the international clinical diabetic retinopathy scale
(11) in this study, and it also related to severity scale in the Early Treatment Diabetic Retinopathy Study (ETDRS) (12). Emergence of microaneurysm and/or blot hemorrhage and/or hard exudate corresponded to
mild nonproliferative diabetic retinopathy (NPDR) and also corresponded to ETDRS level 20, because first findings were microaneurysms
in all but one patient (2). The laser photocoagulation performed on
nonperfused area recognized by fluorescein angiogram after the identification of soft exudates and/or intraretinal microvascular abnormalities
corresponded to severe NPDR, which also corresponded to ETDRS level
53. The presence of a new vessel despite the laser photocoagulation
corresponded to proliferative diabetic retinopathy (PDR), which also
corresponded to ETDRS level 65. Only findings ascertained at two consecutive fundus examinations were judged as positive to lessen potential
errors. The mean ⫾ SD number of eye examinations per patient was 21 ⫾
17 (range, 2–90). In 165 patients, the last fundus observations were
performed from 2003 to 2006. Among the other 71 patients, 22 died
during follow-up, and 49 underwent their last fundus observation at an
earlier time (2000 –2003 in 11 patients, 1995–1999 in 17 patients,
1990 –1994 in 29 patients, and 1983–1989 in 14 patients).
Assessment of long-term glycemic control
Routine measurement of glycosylated hemoglobin (HbA1c) was
started in 1984 in Toranomon Hospital. HbA1c was measured by a chromatographic method (13) with a normal range of 4.8 – 6.1% until May
1997, after which it was measured by a HPLC method with a normal
range of 4.3–5.8% (14). Data obtained with the former method (X) were
converted to match the data obtained with the latter method (Y) by using
the following formula: Y ⫽ 0.918X ⫺ 0.4 (r ⫽ 0.986; n ⫽ 581). Sixteen
patients had no HbA1c data. For the other 238 patients, the mean HbA1c
value during the entire illness was determined as follows: mean HbA1c
values were calculated for each 5-yr period after the onset of diabetes,
and these values were further averaged to exclude the influence of differences in the density of measurements between the periods. The number of HbA1c measurements during each 5-yr period ranged from 2 to 88
(median, 22). In 137 patients, HbA1c data were available for the entire
disease duration (range, 2–27 yr; median, 12 yr). In the other 101 patients, some data were missing; among these, 73 data points were missing
before 1984. The number of patients examined for ␤-cell function, retinopathy findings, and long-term glycemic control and their relations
were illustrated in Fig. 1.
Both systolic blood pressure persistently above 140 mm Hg and diastolic pressure persistently above 90 mm Hg at consecutive determinations or use of an antihypertensive drug was regarded as the evidence of
hypertension, as described previously (2).
Human leukocyte antigen (HLA) typing
HLA-DR and -DQ alleles were typed by previously described PCRrestriction fragment length polymorphism methods (6). HLA-A alleles
were typed by the microcytotoxicity test or the PCR-restriction fragment
length polymorphism method (6).
Definitions of type 1 diabetes subtypes
According to the time from the diagnosis of diabetes to the start of
insulin therapy, the patients were divided into a group with acute-onset
(⬍12 months) type 1 diabetes (n ⫽ 142) and a group with slow-onset
(⬎12 months) type 1 diabetes (n ⫽ 99) (6, 15, 16). Patients who developed ketoacidosis within 1 wk of the onset of hyperglycemic symptoms
and had a concomitant rise of pancreatic exocrine enzymes were classified as having fulminant type 1 diabetes (n ⫽ 10) (6, 17).
J Clin Endocrinol Metab, December 2008, 93(12):4759 – 4766
FIG. 1. Number of patients examined for ␤-cell function (C-peptide
measurement), retinopathy findings, and long-term glycemic control (HbA1c
data) and their relations. Plus (⫹) or minus (⫺) indicate presence or absence of
the data, respectively.
Statistical analysis
The Kaplan-Meier method was used to estimate the cumulative incidence of each type of diabetic retinopathy, and differences between
incidence curves were assessed by the log-rank test (18). Incidence rates
were expressed as the number of events per 100 patient-years, calculated
as the ratio of the observed number of events to the total number of
patient-years of exposure. Cox’s proportional hazards model (18) was
used to examine the combined influence of the variables on the risk of the
development of diabetic retinopathy. The Mann-Whitney U test was
used to compare unpaired data. Differences of frequency between two
groups were assessed by Fisher’s exact probability test. Results are expressed as the means ⫾ SD. All analyses were performed with the JMP6.0
statistical package (SAS Institute Japan, Tokyo, Japan).
Results
Time course of ␤-cell destruction and stratification of
the patients
Among 211 patients for whom longitudinal data on residual
␤-cell function were available, ␤-cell function became undetect-
jcem.endojournals.org
4761
able in 81 patients, and 93% (75 of 81) of these events occurred
within 20 yr after the onset of diabetes (Fig. 2A). On the other
hand, among 130 patients who had residual ␤-cell function
throughout their observation periods, 37 patients still displayed
some ␤-cell function even 20 yr after the onset of diabetes (Fig.
2B). Adding seven patients in whom loss of detectable ␤-cell
function was observed more than 20 yr after the onset of diabetes
(Fig. 2A) to these 37 patients. 20% (44 of 221) of all patients had
residual ␤-cell function even 20 yr after the onset. Accordingly,
we first stratified the patients into a group showing loss of detectable ␤-cell function within 20 yr of the onset of diabetes and
a group with residual ␤-cell function at 20 yr (Table 1). To examine the influence of the duration of residual ␤-cell function on
the onset and progression of diabetic retinopathy, we next stratified the patients by 5-yr intervals stepwise from 20 yr (Table 1).
In these analyses, 30 patients who showed undetectable ␤-cell
function at more than 5 yr after the onset of diabetes by the first
assessment were also included as those without detectable ␤-cell
function in the appropriate strata if the assessment time was
earlier than each designated time point (Table 1).
Stratified analysis of retinopathy
Mild NPDR, severe NPDR, and PDR were detected in 129,
90, and 46 patients during 2708, 3405, and 4021 patient-years
of observation, respectively. Incidence rate of each type of retinopathy according to the status of ␤-cell function at each designated time point was shown in Table 1. The cumulative incidence of mild NPDR was higher among the patients who showed
undetectable ␤-cell function within 20, 15, and 10 yr after the
onset of diabetes than among those with residual ␤-cell function
after the corresponding periods (P ⫽ 0.013, P ⫽ 0.006, and P ⫽
0.048, respectively; Fig. 3D, C, and B). In contrast, the cumulative incidence of mild NPDR did not differ between those who
lost detectable ␤-cell function within 5 yr of the onset of diabetes
and those with residual ␤-cell function at 5 yr after the onset (P ⫽
0.84; Fig. 3A).
Severe NPDR was found earlier in the patients who showed
undetectable ␤-cell function within 20 yr of the onset of diabetes than in those with residual ␤-cell function at 20 yr (P ⫽
0.031). Cumulative incidence of severe NPDR reached 50% at
19.2 yr after onset of diabetes in the former and at 25.8 yr after
FIG. 2. Histograms of patients who showed undetectable ␤-cell function (A) and detectable ␤-cell function (B) at final assessment of C-peptide during follow-up of 211
patients. The patients with residual ␤-cell function at the designated time point consisted of those who showed detectable ␤-cell function after that time and those
who showed undetectable ␤-cell function after that time in the longitudinal observation. The patients without detectable ␤-cell function at the designated time point
consisted of those who showed undetectable ␤-cell function until that time in the longitudinal observation, and those who showed undetectable ␤-cell function at
more than 5 yr after the onset of diabetes by the first assessment of ␤-cell function if the assessment time was earlier than the designated time point, the number of
whom was shown in parentheses in Table 1.
50.3 (72/143)
43.6 (65/149)
68.6 (24/35)
70.3 (26/37)
4.12
2.36
0.99
18.0 (25/139)
40.0 (14/35)
3.44
1.96
0.77
8.19 ⫾ 1.10
17.9 (27/151)
8.29 ⫾ 1.09
8.1 (3/37)
ⴙ
152
84/68
34 ⫾ 14
1981 ⫾ 11
ⴚ
37
23/13
40 ⫾ 13
1991 ⫾ 6
0.0054
0.060
0.011
0.76
0.21
0.36
0.027
⬍0.0001
P
ⴚ
4.77
2.24
1.08
63.3 (38/60)
66.1 (37/56)
38.2 (21/55)
8.42 ⫾ 1.00
6.7 (4/60)
60 (6)
40/20
38 ⫾ 14
1988 ⫾ 7
ⴙ
4.15
2.39
1.00
45.6 (47/103)
46.5 (46/99)
14.6 (14/96)
8.18 ⫾ 1.09
22.1 (23/104)
105
63/42
33 ⫾ 13
1978 ⫾ 10
0.035
0.038
0.0013
0.26
0.015
0.41
0.014
⬍0.0001
P
5.22
2.14
0.76
65.5 (55/84)
69.7 (53/76)
41.3 (31/75)
8.54 ⫾ 1.03
8.3 (7/84)
84 (18)
53/31
36 ⫾ 15
1987 ⫾ 8
ⴚ
4.18
2.41
1.02
47.3 (35/74)
47.2 (34/72)
14.5 (10/69)
8.10 ⫾ 1.02
22.7 (17/75)
76
42/34
33 ⫾ 14
1975 ⫾ 10
ⴙ
15 yr
Residual ␤-cell function
0.025
0.0074
0.0004
0.042
0.015
0.34
0.13
⬍0.0001
P
5.41
2.61
1.01
63.5 (66/104)
68.4 (65/95)
36.2 (34/94)
8.46 ⫾ 1.00
14.4 (15/104)
104 (29)
65/39
36 ⫾ 15
1985 ⫾ 9
ⴚ
4.25
2.60
1.16
41.5 (17/41)
43.6 (17/39)
10.8 (4/37)
7.87 ⫾ 0.73
26.8 (11/41)
42
22/20
30 ⫾ 12
1970 ⫾ 10
ⴙ
20 yr
Residual ␤-cell function
0.016
0.0089
0.0049
0.030
0.09
0.27
0.0499
⬍0.0001
P
a
Numbers in parentheses represent the number of the patients who showed undetectable ␤-cell function at more than 5 yr after the onset of diabetes by the first assessment of residual ␤-cell function.
␤-Cell Destruction and Diabetic Retinopathy
Times judging residual ␤-cell function are expressed from the onset of diabetes. Residual ␤-cell function ⫺ or ⫹ means loss of detectable ␤-cell function at the designated time point or residual ␤-cell function at the designated
point.
No. of patientsa
Men/women
Age at onset (yr)
Calendar year of
onset
Mean HbA1c (%)
Frequency of
hypertension
Frequency of 3-allele
combination
(HLA-A24,
-DQA1*03, and
-DR9) (%)
Frequency of 2-allele
combination
(HLA-A24 and
DQA1*03) (%)
Frequency of acuteonset type 1
diabetes (%)
Incidence rate (per
100 patientyears)
Mild NPDR
Severe NPDR
PDR
10 yr
Residual ␤-cell function
Nakanishi and Watanabe
5 yr
Residual ␤-cell function
Time points of judgment regarding the status of residual ␤-cell function
TABLE 1. Clinical characteristics and incidence rate of each type of retinopathy in the patients without detectable ␤-cell function or with residual ␤-cell function at each
designated time point
4762
J Clin Endocrinol Metab, December 2008, 93(12):4759 – 4766
J Clin Endocrinol Metab, December 2008, 93(12):4759 – 4766
jcem.endojournals.org
4763
FIG. 3. Cumulative incidence of mild NPDR in the patients who lost detectable ␤-cell function (solid lines) and those with residual ␤-cell function (broken lines) at 5 yr
(A), 10 yr (B), 15 yr (C), and 20 yr (D) after the onset of diabetes. The Kaplan-Meier data were drawn using 10 patients as the maximum follow-up number.
onset of diabetes in the latter. On the other hand, the cumulative incidence of severe NPDR did not differ between the
patients who showed undetectable ␤-cell function within 15,
10, and 5 yr after the onset of diabetes and those with residual
␤-cell function after the corresponding periods (data not
shown).
Although the cumulative incidence of PDR was higher in the
patients who lost detectable ␤-cell function within 10 yr of the onset
of diabetes, in whom it reached 50% at 24.7 yr after the onset of
diabetes, than in those who retained ␤-cell function after 10 yr,
in whom it reached 50% at 34.0 yr after the onset of diabetes (P ⫽
0.01), it did not differ between those showing undetectable ␤-cell
function within 20, 15, or 5 yr after the onset of diabetes and
those who had residual ␤-cell function after the corresponding
periods (data not shown).
Stratified analysis of glycemic control
The mean HbA1c value was higher in the patients who showed
undetectable ␤-cell function within 20 or 15 yr from the onset of
diabetes than in those with residual ␤-cell function after the corresponding periods, whereas it did not differ between those who
lost detectable ␤-cell function within 10 or 5 yr from the onset of
diabetes and those who retained ␤-cell function after the corresponding periods (Table 1).
Genetical and clinical characteristics of the patient strata
Because ␤-cell status affected the development of mild NPDR
most strongly, genetic and clinical characteristics of the patient
strata were analyzed in relation to mild NPDR.
The frequencies of a 3-HLA allele combination (HLA-A24,
-DQA1*03, and -DR9) and a 2-HLA allele combination (HLAA24 and -DQA1*03), which are known to be associated with the
acute onset of type 1 diabetes and early complete ␤-cell destruction (6), were higher in the patients who showed undetectable
␤-cell function at any time point during the study than in the
patients who retained ␤-cell function after the corresponding
periods, except for the combination of HLA-A24 and
-DQA1*03 at 5 yr after the onset of diabetes (Table 1). The
cumulative incidence of mild NPDR was higher in patients with
the 2-allele combination (n ⫽ 125), in whom it reached 50% at
14.7 yr after onset of diabetes, than in those without this combination (n ⫽ 81), in whom it reached 50% at 18.5 yr after onset
of diabetes (P ⫽ 0.041), although it did not differ between patients with the 3-allele combination (n ⫽ 55) and those without
this combination (n ⫽ 147) (data not shown).
Acute-onset type 1 diabetes was more common among the patients who showed undetectable ␤-cell function at any time point
during the study than in those who retained ␤-cell function after the
corresponding periods (Table 1). The cumulative incidence of mild
NPDR was higher in the patients with acute-onset diabetes (n ⫽
131), in whom it reached 50% at 12.2 yr after the onset of diabetes,
than in those with slow-onset diabetes (n ⫽ 86), in whom it reached
50% at 14.6 yr after the onset of diabetes (P ⫽ 0.044).
The ages at the onset of diabetes were older, frequency of
hypertension was lower, and the calendar year of onset was later
in the patients without ␤-cell function than those with residual
␤-cell function at the time points of 20, 10, and 5 yr, 15 and 10
␤-Cell Destruction and Diabetic Retinopathy
0.47
0.14
0.69
0.89 (0.65–1.22)
0.98 (0.96 –1.01)
1.00 (0.99 –1.01)
0.25
0.62
0.73
0.85 (0.65–1.12)
1.00 (0.98 –1.01)
1.00 (0.99 –1.00)
Time point of judgment of residual ␤-cell function is expressed as the duration from the onset of diabetes.
0.12
0.27
0.60
0.81 (0.62–1.05)
1.01 (0.99 –1.03)
1.00 (0.99 –1.00)
0.89 (0.68 –1.16)
1.00 (0.98 –1.02)
1.00 (0.99 –1.01)
0.37
0.84
0.37
0.87
0.74
0.012
0.37
1.75 (1.14 –2.57)
1.16 (0.84 –1.63)
0.079
0.46
1.47 (0.95–2.13)
1.11 (0.84 –1.50)
0.49
0.61
P
1.19 (0.81–1.67)
1.03 (0.79 –1.35)
1.14 (0.76 –1.63)
1.07 (0.82–1.42)
1.94 (1.36 –2.81)
1.63 (1.04 –2.72)
⬍0.0001
0.0048
1.75 (1.33–2.33)
1.54 (1.14 –2.10)
1.96 (1.48 –2.60)
1.43 (1.07–1.90)
⬍0.0001
0.40
1.88 (1.42–2.51)
1.19 (0.77–1.73)
⬍0.0001
0.016
20 yr
HR (95% CI)
P
15 yr
HR (95% CI)
P
10 yr
HR (95% CI)
The simplest way to determine the period of residual ␤-cell function that is sufficient to protect patients with type 1 diabetes
against the early occurrence of diabetic retinopathy would be to
compare the incidence of retinopathy among patient groups who
lost detectable ␤-cell function within various given periods, for
example, at 5-yr intervals. However, it is difficult to achieve this
in a follow-up study because there are always censored cases.
Therefore, we chose to compare the incidence between groups
whose ␤-cell status was judged at designated time points set at
5-yr intervals. If the difference of the incidence between stratified
groups was significantly affected by a stepwise 5-yr shift of the
assessment time point, then the 5-yr period in question could be
assumed to be pivotal in determining the early occurrence of
retinopathy.
This study showed that retaining some ␤-cell function for
more than 20, 15, or 10 yr after the onset of diabetes delayed the
development of mild NPDR, and preservation of residual ␤-cell
function for more than 20 or 15 yr after the onset of diabetes was
associated with lower mean HbA1c values. The most significant
risk factor for mild NPDR was mean HbA1c values in multivariate analysis, however. These findings suggest that the better
glycemic control throughout the entire duration of disease attained by preserved ␤-cell function for at least 10 yr protects
patients with type 1 diabetes from the early development of diabetic retinopathy. Glycemic exposure for a long period is established as a primary risk factor for retinopathy in the large
cohort with type 1 (19) as well as type 2 diabetes (20).
Time points of judgment of residual ␤-cell function
Discussion
TABLE 2. Relative hazard of the development of diabetic retinopathy according to Cox’s proportional hazards model in the patients with type 1 diabetes
Multivariate analysis
To examine the influence of loss of detectable ␤-cell function
until various designated time points on the development of diabetic retinopathy after adjusting for possible confounders, hazard ratios (HRs) and the associated 95% confidence intervals
(CIs) were estimated in each data set using Cox’s proportional
hazards model (Table 2). In this model, the response variable was
the time until the occurrence of each type of retinopathy or the
time until the last eye examination. The covariates examined
included the gender, hypertension, and possession of HLA-A24
and -DQA1*03 as categorical variables, as well as the mean
HbA1c value, the period before insulin therapy, and the age at
onset as continuous variables (Table 2).
Loss of detectable ␤-cell function until 20, 15, or 10 yr from
the onset of diabetes was shown to be an independent risk factor
for the development of diabetic retinopathy (Table 2). In contrast, loss of detectable ␤-cell function by 5 yr after the onset of
diabetes was not an independent risk factor for diabetic retinopathy (Table 2). Mean HbA1c value was a most significant risk
factor for the development of diabetic retinopathy in any data set
(Table 2). In the data set for residual ␤-cell function at 20 yr after
the onset of diabetes, hypertension was an independent risk factor for the development of diabetic retinopathy (Table 2).
Mean HbA1c (%)
Detectable ␤-cell function (0 ⫽ present, 1
⫽ absent)
Hypertension (0 ⫽ no, 1 ⫽ yes)
Possession of HLA-A24 and -DQA1*03
(0 ⫽ no, 1 ⫽ yes)
Sex (0 ⫽ women, 1 ⫽ man)
Age at onset (yr)
Period before insulin therapy (months)
yr, and every time point after the onset of diabetes, respectively
(Table 1).
0.0002
0.034
J Clin Endocrinol Metab, December 2008, 93(12):4759 – 4766
P
Nakanishi and Watanabe
5 yr
HR (95% CI)
4764
J Clin Endocrinol Metab, December 2008, 93(12):4759 – 4766
On the other hand, after adjusting for possible confounders
including the mean HbA1c value, multivariate analysis showed
that loss of detectable ␤-cell function by 20, 15, or 10 yr after the
onset of diabetes was a risk factor for the development of diabetic
retinopathy. It suggests that, even if the mean HbA1c is the same,
better stability of blood glucose achieved by residual ␤-cell function
(7) may be protective against diabetic retinopathy because glycemic
spikes are suspected to be a risk factor for microangiopathy (21).
The influence of physiological effects of C-peptide (22) on the protection of diabetic retinopathy is uncertain from this study.
Acceleration of progression to severe NPDR or PDR by loss
of detectable ␤-cell function was less marked than the influence
on the development of mild NPDR in this study. This may have
been partly due to the lower event rates of severe NPDR and PDR
compared with mild NPDR, which meant that our study had
limited statistical power to detect differences. Alternatively, although the extent of ␤-cell destruction obviously influences glycemic control, progression to proliferative retinopathy may also
be influenced by other variables such as production of vascular
endothelial growth factor (23).
The combination of HLA-A24 and -DQA1*03 is known as a
marker for the acute onset of type 1 diabetes and early complete
␤-cell destruction (6), and it was a surrogate marker for early
development of diabetic retinopathy in this study. In addition,
the acute onset of type 1 diabetes, which reflects a higher rate of
␤-cell destruction (5) and is strongly associated with early complete ␤-cell loss (6), was also a surrogate marker for the early
development of diabetic retinopathy. These surrogate markers
are correlated with diabetic retinopathy indirectly through ␤-cell
destruction and subsequent worsening of glycemic control. An
older age for disease onset was also associated with earlier complete ␤-cell loss, which may have been due to including patients
with fulminant type 1 diabetes that develops later than acuteonset type 1 diabetes (24). However, an older age of onset was
not associated with any type of diabetic retinopathy. The higher
age of onset in our patients compared with Caucasians may also
be due to the inclusion of patients with slow-onset type 1 diabetes, which also develops later than acute-onset type 1 diabetes
(16). In the data set for assessing ␤-cell function at 20 yr after the
onset of diabetes, hypertension was an independent risk factor
for mild NPDR as described (25). The calendar year of the onset
of diabetes was later in the patients without residual ␤-cell function at any time point, which was thought to be due to exclusion
of the patients who had no C-peptide data during the first 5 yr
and showed undetectable ␤-cell function thereafter from the longitudinal study because such patients were prevalent in earlier
years in this study.
This study was performed retrospectively, and the observations were not standardized. These are limitations with respect to
delineating longitudinal changes of residual ␤-cell function (6)
and retinopathy findings. Missing HbA1c data including those
before 1984 are also limitations for assessment of long-term glycemic control. Limitations also exist in the method of grading retinopathy in this study. Indirect ophthalmoscopy by one person may
be less reliable and more prone to bias than stereoscopic fundus
photographs assessed by independent observers. Prospective study
jcem.endojournals.org
4765
on C-peptide and retinopathy findings using stereoscopic fundus
photographs will be needed to confirm our study.
Acknowledgments
We thank Fumie Takano for her secretarial work.
Address all correspondence and requests for reprints to: Koji Nakanishi,
Department of General Internal Medicine and Metabolism, Toranomon
Hospital, 1-3-1 Kajigaya, Takatsu-ku, Kawasaki, Kanagawa 213-8587,
Japan. E-mail: [email protected].
This work was supported by the Japanese Ministry of Education,
Science, and Culture (Grant 14571117).
Disclosure Statement: The authors have nothing to disclose.
References
1. Palmer JP, Fleming GA, Greenbaum CJ, Herold KC, Jansa LD, Kolb H, Lachin
JM, Polonsky KS, Pozzilli P, Skyler JS, Steffes MW 2004 C-peptide is the
appropriate outcome measure for type 1 diabetes clinical trials to preserve
␤-cell function. Report of an ADA workshop, 21–22 October 2001. Diabetes
53:250 –264
2. Nakanishi K, Kobayashi T, Inoko H, Tsuji K, Murase T, Kosaka K 1995
Residual ␤-cell function and HLA-A24 in IDDM. Markers of glycemic control
and subsequent development of diabetic retinopathy. Diabetes 44:1334 –1339
3. The Diabetes Control and Complications Trial Research Group 1998 Effect of
intensive therapy on residual ␤-cell function in patients with type 1 diabetes in
the Diabetes Control and Complications Trial. A randomized, controlled trial.
Ann Intern Med 128:517–523
4. Steffes MW, Jackson M, Sibley S, Thomas W 2003 ␤-Cell function and the
development of diabetes-related complications in the Diabetes Control and
Complications Trial. Diabetes Care 26:832– 836
5. American Diabetes Association 2005 Diagnosis and classification of diabetes
mellitus. Diabetes Care 28(Suppl 1):S37–S42
6. Nakanishi K, Inoko H 2006 Combination of HLA-A24, -DQA1*03, and -DR9
contributes to acute onset and early complete ␤-cell destruction in type 1 diabetes:
longitudinal study of residual ␤-cell function. Diabetes 55:1862–1868
7. Nakanishi K, Kobayashi T, Miyashita H, Ohkubo M, Sugimoto T, Murase T,
Kosaka K, Inouye K, Kono M 1990 Relationships among islet cell antibodies,
residual ␤-cell function, and metabolic control in patients with insulin-dependent diabetes mellitus of long duration: use of a sensitive C-peptide radioimmunoassay. Metabolism 39:925–930
8. Nakanishi K, Kobayashi T, Murase T, Nakatsuji T, Inoko H, Tsuji K, Kosaka
K 1993 Association of HLA-A24 with complete ␤-cell destruction in IDDM.
Diabetes 42:1086 –1093
9. Aiello LM, Rand LI, Sebestyen JG, Weiss JN, Bradbury MJ, Wafari MZ,
Briones JC 1985 The eyes and diabetes. In: Marble A, Krall LP, Bradley RF,
Christlieb AR, Soeldner JS, eds. Joslin’s diabetes mellitus. 12th ed. Philadelphia: Lea and Febiger; 600 – 634
10. Matsui M, Sato Y 1989 Classification of diabetic retinopathy and application
standard of photocoagulation. Ophthalmol Jpn 93:803– 808
11. Wilkinson CP, Ferris III FL, Klein RE, Lee PP, Agardh CD, Davis M, Dills D,
Kampik A, Pararajasegaram R, Verdaguer JT 2003 Proposed international
clinical diabetic retinopathy and diabetic macular edema disease severity
scales. Ophthalmology 110:1677–1682
12. Early Treatment Diabetic Retinopathy Study Research Group 1991 Fundus
photographic risk factors for progression of diabetic retinopathy. ETDRS report no. 12. Ophthalmology 98:823– 833
13. Wajcman H, Dastugue B, Labie D 1979 Quantitation of hemoglobin A1c: a
rapid, automated precision chromatography technique. Clin Chim Acta 92:
33–39
14. Shima K, Endo J, Oimomi M, Omori Y, Katayama Y, Kanazawa Y, Kawai T,
Kawamori R, Kanno T, Kiyose H, Kuwajima M, Nakashima K, Nagamine Y,
Baba S, Hoshino T 1998 Interlaboratory difference in GHb measurement in
Japan. The fifth report of the GHb standardization committee, the Japan Diabetes Society. J Japan Diab Soc 41:317–323
15. Nakanishi K, Kobayashi T, Miyashita H, Okubo M, Sugimoto T, Murase T,
Hashimoto M, Fukuchi S, Kosaka K 1994 Exocrine pancreatic ductograms in
insulin-dependent diabetes mellitus. Am J Gastroenterol 89:762–766
4766
Nakanishi and Watanabe
␤-Cell Destruction and Diabetic Retinopathy
16. Kobayashi T, Tamemoto K, Nakanishi K, Kato N, Okubo M, Kajio H, Sugimoto
T, Murase T, Kosaka K 1993 Immunogenetic and clinical characterization of
slowly progressive IDDM. Diabetes Care 16:780 –788
17. Imagawa A, Hanafusa T, Miyagawa J, Matsuzawa Y 2000 A novel subtype of
type 1 diabetes mellitus characterized by a rapid onset and an absence of
diabetes-related antibodies. N Engl J Med 342:301–307
18. Cox DR, Oakes D 1984 Analysis of survival data. London: Chapman and Hall;
91–111
19. Lachin JM, Genuth S, Nathan DM, Zinman B, Rutledge BN 2008 Effect of
glycemic exposure on the risk of microvascular complications in the diabetes
control and complications trial–revisited. Diabetes 57:995–1001
20. Stratton IM, Kohner EM, Aldington SJ, Turner RC, Holman RR, Manley SE,
Matthews DR 2001 UKPDS 50: risk factors for incidence and progression of
retinopathy in type II diabetes over 6 years from diagnosis. Diabetologia 44:
156 –163
J Clin Endocrinol Metab, December 2008, 93(12):4759 – 4766
21. Ceriello A 2003 The possible role of postprandial hyperglycaemia in the pathogenesis of diabetic complications. Diabetologia 46(Suppl 1):M9 –M16
22. Wahren J, Ekberg K, Jörnvall H 2007 C-peptide is a bioactive peptide. Diabetologia 50:503–509
23. Al-Kateb H, Mirea L, Xie X, Sun L, Liu M, Chen H, Bull SB, Boright AP,
Paterson A 2007 Multiple variants in vascular endothelial growth factor
(VEGFA) are risk factors for time to severe retinopathy in type 1 diabetes. The
DCCT/EDIC genetics study. Diabetes 56:2161–2168
24. Imagawa A, Hanafusa T, Uchigata Y, Kanatsuka A, Kawasaki E, Kobayashi
T, Shimada A, Shimizu I, Toyoda T, Maruyama T, Makino H 2003 Fulminant
type 1 diabetes: a nationwide survey in Japan. Diabetes Care 26:2345–2352
25. Krolewski AS, Warram J 2005 Epidemiology of late complications of diabetes: a
basis for the development and evaluation of preventive program. In: Kahn CR,
Weir GC, King GL, Jacobson AM, Moses AC, Smith RJ, eds. Joslin’s diabetes
mellitus. 14th ed. Philadelphia: Lippincott Williams, Wilkins; 795– 808