Type 1 diabetes duration decreases pulmonary diffusing capacity during exercise

Abstract

Background

Diabetes damages peripheral tissues, however, its effects on the lung are less known. Lung diffusing capacity (D_LCO) is influenced by alveolar-capillary membrane conductance (D_M) and pulmonary capillary blood volume (V_C), both of which are reduced in adults with type-1 diabetes (T1D).

Objective

We sought to determine if diabetes duration affects D_LCO, D_M, V_C, and cardiac output (Q).

Methods

24 type-1 diabetics (10.7 – 52.8 years) and 24 non-diabetic controls were recruited and had D_LCO, D_M, V_C, and Q measured at rest and exercise (40, 70 and 90% VO₂max).

Results

When stratified into two groups based on age (young < 20.6 years old), there were no significant differences in D_LCO, D_M, V_C, or Q (all of which were normalized to body surface area [BSA]) in the young group, or in the old group. When stratified by diabetes duration (short duration 0.33 – 8.9 years, long duration 9.6 – 28 years), the T1D in the long duration group had lower D_LCO/BSA and D_M/BSA compared to the controls (P<0.05). There were no differences in any of the variables in the short duration group.

Conclusions

This study has shown that duration of diabetes is associated with decrements in diffusing capacity and its components.

Introduction

The American Diabetes Association recommends that people with type-1 diabetes (T1D) engage in physical activity [1, 2]. It is unclear whether the effects on the lung limit the increasing number of people with diabetes who are participating in ‘adventure’ activities at moderate to high altitude [3]. Adults with type 1 diabetes have pulmonary diffusion limitations [4, 5] that are particularly applicable when increased altitude reduces the alveolar-arterial gradient for oxygen transfer [6, 7]. Exercise exacerbates this challenge by increasing cardiac output and reducing the capillary transit times available for oxygen transfer in the lung [8]. Consequently, adventure activities at moderate to high altitude may result in more arterial hypoxemia, limiting the ability of individuals with diabetes to sustain vigorous activities at increased altitude.

The diffusing capacity of the lung (D_LCO) is determined by two components, the alveolar-capillary membrane diffusing capacity (D_M) and the pulmonary capillary blood volume (V_C) [9]. Diabetes thickens the pulmonary capillary basal lamina [10, 11] and increases endothelial permeability [12], reducing D_M. Niranjan et al. [13] showed that during exercise D_LCO was lower in healthy type 1 diabetic adults and that the reduction in D_LCO was attributable to reduced D_M. They also showed that five years of intensive glycemic control prevented further deterioration of D_LCO. It is unclear whether adolescents with diabetes have the same impairment in pulmonary diffusion as adults. Van Gent et al. [14] found that D_LCO/V_A was not different from age-dependent norms in 27 children with T1D. In contrast, Cazzato et al. [15] found a lower D_LCO in children with T1D than in their age-matched controls but found no correlation with diabetes duration or HbA1c. However Villa et al. [16] found D_LCO was reduced in adolescents with diabetes and D_LCO was inversely associated with HbA1c. These data suggest that age, magnitude, or duration of hyperglycemia impairs pulmonary diffusion by altering the alveolar capillary membrane, but these changes to the diabetic lung are preventable.

Studies in diabetic adults suggest that reduced V_C [13, 17] causes reduced D_LCO. V_C is determined by the number of pulmonary capillaries in contact with ventilated alveoli. During exercise, lung volumes and pulmonary blood flow increase, causing an increase in D_LCO. Exercise is needed to evaluate V_C because only a small percentage of the pulmonary vasculature is used at rest, therefore, resting measures do not adequately assess microvascular reserve. An attenuated increase in cardiac output during exercise is well described in T1D [13, 18], and could reduce V_C by limiting recruitment and distension of pulmonary vessels [19, 20, 21, 22]. Furthermore, microangiopathy in the lung in T1D may impair pulmonary vascular recruitment and distension by not allowing the normal increase in V_C [5].

The purpose of this study was to compare resting and exercising D_LCO, D_M and V_C, in children and adults with and without T1D at 7000 ft (2, 130 meters, barometric pressure ~600 mmHg) to examine the effect of age and duration of disease on lung diffusing capacity and its components. We tested two hypotheses. The first was that older people with T1D would have a lower D_LCO compared to controls, but younger people with T1D would not. The second hypothesis was that subjects with a longer duration of diabetes (independent of age) would have a lower D_LCO compared to controls.

Methods

Subjects

Twenty-four individuals with type-1 diabetes and 24 non-diabetic controls matched for age, self-reported fitness history, and gender participated in this study. Subjects were aged 11 – 53 years, were non-smoking and showed no evidence of cardiovascular disease (Bruce protocol), microvascular impairment (proteinurea or retinopathy) or autonomic dysfunction. Three T1D participants were taking ACE inhibitors, however no participants were hypertensive (>140/90 mm Hg). The protocol was reviewed and approved by the Northern Arizona University Institutional Review Board and all subjects signed a written informed consent prior to the study. In addition, adolescents filled out a self-assessment for Tanner stage of maturity [23].

Procedure

The study required two visits to the lab. During the first visit, height, weight and body fat percentage were determined using a seven-site skinfold method [24] on adults, and a two-site skinfold method on adolescents [25]. Standard spirometry testing was performed using a Medical Graphics CPX/D (St. Paul, MN) metabolic cart. Peak oxygen uptake (VO₂peak) (ParvoMedics TrueOne 2400 Sandy, Utah) was then determined on a stationary cycle ergometer (Corival Lode B.V., The Netherlands).

Peak oxygen uptake was assessed using an incremental protocol. Adolescents pedaled at 60 – 70 revolutions per minute (rpm) at an initial workload of 25 watts for subjects less than 14 years old and 40 watts for those greater than 14 years old. Workload then increased 15 watts each minute until exhaustion. Adults cycled 70 – 80 rpm. Females started at a workload of 50 watts and males at 100 watts and workload increased by 50 watts and then 25 watts every 2 minutes until exhaustion to make the total exercise duration between 8 and 14 minutes. Since the test was performed on a cycle ergometer, classification of peak VO₂ was based on either; 1) an increase in VO₂ of less than 250 mL with an increase in workload, or 2) two of the following three situations; respiratory exchange ratio (RER) greater than 1.10, the subject’s inability to maintain the pedal rate, and a heart rate within 10 beats per minute of age-predicted maximum.

During the second visit, cardiac output (Q), D_LCO, and lung diffusing capacity for nitric oxide (D_LNO) were measured using a rebreathing method. Two maneuvers were performed at rest and during each of the three exercise workloads (40%, 70%, and 90% of VO₂peak) for the measurement of D_LCO and D_LNO, and the determination of D_M [26], while simultaneously measuring Q. Subjects rebreathed a gas mixture containing 0.6% C₂H₂, 0.28% C¹⁸O, 9% He, 36.1% O₂, and 40 ppm NO from a 5-liter anesthesia bag for eight breaths. Breathing was set at a rate of 32 breaths/min using a metronome until respiratory rate exceeded 32 breaths/min, at which point the metronome was set at the subject’s preferred breathing rate. Gases were sampled with a Perkin Elmer MGA-1100 mass spectrometer (Wesley, MA) and Sievers 280i Nitric Oxide Analyzer (GE, Boulder, CO), and analyzed on a PC using custom software (KC Beck, Liberty, UT) for the measurement of D_LCO, D_LNO, and Q. The bag volume corresponded to the subject’s normal tidal volume plus 200 mL of the gas mixture. During rebreathing, D_LCO, D_LNO, and Q were measured from the slope of the end-tidal exponential disappearance of C¹⁸O, NO, and acetylene respectively, with respect to helium over time. These data were used to calculate the two components of D_LCO, D_M and V_C, using the Roughton and Forster [9] equation.

Statistical Analysis

Data are reported as mean ± SD. Statistical analyses were performed using SigmaStat 3.5 (Systat Software, Inc., Point Richmond, CA). A two-way analysis of variance with repeated measures was used to determine differences in Q, D_LCO, D_M, and V_C (dependent variables) between groups, exercise intensity, and the interaction of group and condition (independent variables). The post-hoc test, Holm-Sidak, was used when appropriate. Results were considered statistically significant if p< 0.05.

Results

Subject testing starting in February 2011 and was completed in September 2012. Whole group subject characteristics are shown in table 1. The T1D and control groups were of similar age, height, weight, percent body fat, and activity level. There were no group differences in VO₂peak, RER and heart rate at maximal exercise, or pulmonary function. D_LCO/BSA, D_M/BSA, V_C/BSA, and Q/BSA were measured at rest, and low, moderate, and high exercise intensities. D_LCO/BSA, D_M/BSA, and Q/BSA increased with increased exercise intensity. D_LCO/BSA (p = 0.167), D_M/BSA (p = 0.494), V_C/BSA (p = 0.569), or Q/BSA (p = 0.112) were not different across the testing conditions (Table 2). Relative exercise intensities were not different between groups (p = 0.390).

Table 1.

Subject Characteristics, Data are means ± SD

	Diabetics	Controls	p-value
Age	23.8±12.2	24.2±12.9	0.238
Gender (% male)	45.8	45.8	NA
Height (cm)	166.6±14.7	166.8±12.2	0.936
Weight (kg)	64.1±18.5	65.4±20.2	0.593
BMI	22.5±3.3	23.0±4.3	0.482
Body Fat Percentage	23.8±	22.9±7.0	0.463
Activity Level (hrs/week)	5.0±2.8	6.0±3.7	0.139
Diabetes Duration (yrs)	9.8±7.8	----	----
HbA1c	7.9±1.3 (n=18)	----	----
VO2max (mL/kg/min)	34.9±5.8	36.3±7.7	0.307
Max Heart Rate (bpm)	181±13.8	187.0±14.1	0.136
RER at max	1.1±0.1	1.1±0.1	0.951
FVC (% predicted)	100.4±12.1	104.9±11.3	0.146
FEV1.0 (% predicted)	97.6±15.9	100.9±11.6	0.379

Table 2.

Resting and Exercise Variables Measured in Diabetics and their Matched Controls, Data are means ± SD

	Diabetics				Controls
Exercise Intensity	Rest	Low	Moderate	High	Rest	Low	Moderate	High
% of VO2max	15.6±3.1	43.4±3.9	72.2±4.6	92.5±6.0	15.7±4.1	43.3±5.9	70.3±8.7	89.3±10.4
DLCO/BSA	12.9±2.5	15.7±2.7	17.9±3.4	19.5±4.3	13.4±2.2	16.3±3.5	19.5±3.8	21.5±4.6
DM/BSA	18.1±5.4	20.8±3.6	23.9±4.9	28.2±7.0	18.4±4.8	21.4±5.3	25.9±6.3	28.8±6.3
VC/BSA	56.7±70.9	64.2±40.2	73.3±59.9	60.8±24.9	56.9±30.2	68.8±30.8	73.2±30.6	75.4±21.5
Q/BSA	3.1±0.6	5.9±1.3	7.4±1.4	8.6±1.7	3.2±0.8	6.2±1.4	8.3±1.5	9.5±1.5

The effect of age on lung diffusing capacity was determined by stratifying subjects around the median age (20.7 years) of the sample. The average age of the young and older groups were 15.3 ± 4.0 years and 32.3 ± 11.8 years respectively (P< 0.05). In the young group, D_LCO/BSA, D_M/BSA, and Q/BSA, but not Vc/BSA increased with increasing exercise intensity. In the older group, D_LCO/BSA, D_M/BSA, V_C/BSA, and Q/BSA increased with increasing exercise intensity. There were no differences in D_LCO/BSA, D_M/BSA, V_C/BSA, and Q/BSA between the young and older T1Ds and their age-matched controls (table 3).

Table 3.

Resting and Exercise Variables in Diabetics and Controls Stratified by Age, Data are means ± SD

	Young								Older
	Diabetics				Controls				Diabetics				Controls
Exercise Intensity	Rest	Low	Moderate	High	Rest	Low	Moderate	High	Rest	Low	Moderate	High	Rest	Low	Moderate	High
DLCO/BSA	13.1±3.2	15.7±3.5	17.4±3.9	19.4±5.0	13.7±1.9	15.6±3.8	19.6±3.9	20.9±4.9	12.9±2.0	15.7±1.8	18.3±3.0	19.7±3.7	13.0±2.5	17.0±3.3	19.5±3.9	22.0±4.4
DM/BSA	17.2±6.1	20.9±4.6	23.2±6.0	28.5±8.7	19.5±5.0	20.2±4.6	25.8±5.3	27.9±6.2	18.7±4.9	20.7±2.6	24.5±3.7	28.0±5.2	17.5±4.8	22.5±6.0	26.1±7.4	29.6±6.5
VC/BSA	81.8±105.9	61.4±29.0	87.6±83.8	55.2±18.5	64.1±40.2	71.8±38.5	72.4±38.6	71.1±25.4	37.9±10.3	67.1±50.2	60.2±19.8	66.4±29.8	50.9±18.3	65.9±22.1	73.9±23.0	78.9±17.9
Q/BSA	3.3±0.6	6.1±1.2	7.1±1.0	8.4±2.0	3.3±0.6	5.9±1.3	8.5±1.6	9.4±1.5	2.9±0.6	5.8±1.3	7.7±1.7	8.8±1.3	3.1±0.9	6.4±1.4	8.2±1.3	9.5±1.6

The sample was then stratified around the median duration of diabetes (9.3 years) and those above and below the median duration were compared against their assigned age-and sex-matched controls. In both groups, D_LCO/BSA, D_M/BSA, and Q/BSA increased with increasing exercise intensity (P < 0.05). In the long duration group, V_C/BSA, increased with increasing exercise intensity (P < 0.05). There were no differences in D_LCO/BSA, D_M/BSA, V_C/BSA, and Q/BSA (p = 0.328, 0.155, 0.601, and 0.263, respectively) for the short duration T1D (4.0 ± 2.8 years) vs. non-diabetic controls. Similarly, there were no differences in V_C/BSA and Q/BSA (p = 0.115 and 0.234, respectively) between the long duration T1D (15.7 ± 6.6 years) vs. matched controls. However, long duration T1D had lower D_LCO/BSA (p = 0.014) and D_M (p = 0.033) than matched controls (figure 1, figure 2).

Figure 1.

Figure 1A: D_LCO/BSA at rest, low, moderate, and high exercise intensities for the long duration diabetics and their matched controls. * (p < 0.05) signifies a difference between diabetics and controls. † (p < 0.05) different from rest. ‡ (p < 0.05) different from low intensity exercise # (p < 0.05) different from moderate intensity exercise.

Figure 1B: D_LCO/BSA at rest, low, moderate, and high exercise intensities for the short duration diabetics and their matched controls. † (p < 0.05) different from rest. ‡ (p < 0.05) different from low intensity exercise.

Figure 2.

Figure 2A: D_M/BSA at rest, low, moderate, and high exercise intensities for long duration diabetics and their matched controls. * (p < 0.05) Signifies a difference between diabetics and controls. † (p < 0.05) different from rest. ‡ (p < 0.05) different from low intensity exercise # (p < 0.05) different from moderate intensity exercise.

Figure 2B: D_M/BSA at rest, low, moderate, and high exercise intensities for short duration diabetics and their matched controls. † (p < 0.05) different from rest. ‡ (p < 0.05) different from low intensity exercise.

Discussion

Our study of non-smoking adolescents and adults found that resting and peak exercise pulmonary carbon monoxide diffusing capacity was not significantly different between normals vs. type 1 diabetics. Contrary to our hypothesis, neither younger nor older patients were different from their age- and sex-matched controls. However, when participants were stratified by diabetes duration, we found that people with longer type 1 diabetes diagnoses (>10 years) had reduced D_LCO relative to controls, and this difference was exacerbated during high intensity exercise. Simultaneous measurements of D_LCO and D_LNO in our study showed that the reduction in D_LCO was due to lower alveolar-capillary membrane diffusing capacity (D_M), but that pulmonary capillary blood volume was not significantly affected. None of the participants had evidence of microangiopathy, and there was no significant relationship between D_LCO and HbA1c levels. These findings suggest that diabetes-specific impairment of pulmonary diffusion is a progressive process.

Our finding that age-matched D_LCO did not differ between normal and T1D subjects does not agree with some previous reports of impaired resting pulmonary diffusing capacity in adults [4, 17, 27, 28, 29] and adolescents [14, 16, 30] with type 1 diabetes. However, some [17], but not all [16] of these earlier studies also reported abnormal pulmonary function in their subjects, whereas our type 1 diabetic cohort had normal pulmonary function (spirometry). Normal lung volumes and ventilatory flows in our diabetic cohort may have allowed them to transfer CO more effectively whereas the pathology in the previous studies’ cohorts [14, 15, 17, 30] did not. Nonetheless, a study of children with normal spirometry tests has also reported impaired D_LCO [16].

Another possibility is that differences in global diabetes control or behavior between our cohort and previous cohorts may have contributed to differences in D_LCO observations. Several studies reporting lower D_LCO in patients with type 1 diabetes have included nephropathic [29, 31], neuropathic [29] and retinopathic [28, 32] patients. Other cohorts have been composed of patients with higher mean HbA1c values than the present study [30]. Two studies stratified participants with type 1 diabetes into good and poor glycemic control [16, 33] and found that pulmonary diffusion was impaired in those with poor control, but not in those with HbA1c less than 7% (53 mmol). None of the participants in this study had clinically evident microvascular abnormalities, and HbA1c levels were, in relation to previous studies, lower. Considering the association between glycemic control and microvascular complications [34], group differences in D_LCO may have been even more evident if our cohort had worse metabolic control. The average HbA1c in our subjects was 7.9, while in studies showing differences in D_LCO between diabetics and controls [13, 15, 16, 29, 30, 32, 33, 35], only one study had HbA1c values lower than ours [15], while the rest had an average HbA1c in their subjects of 8.98%. Yet another possibility for the lack of difference within the age groups may be the wide range of durations. The diabetic adolescents ranged from 0.33 to 14 years disease duration, while the adults ranged from 6.5 to 28 years.

Our finding that diabetes duration was associated with reduced pulmonary diffusion expands on the findings of Asanuma et al. [4] who found that the decrease in D_LCO was correlated with diabetes duration in adults with T1D. However, retinopathy was also correlated with diabetes duration in their cohort, raising the possibility that microangiopathy, which worsened with diabetes duration, explained the reduction in D_LCO rather than uncomplicated diabetes duration. Villa et al. [16] found that glycemic control, but not diabetes duration reduced D_LCO in a small cohort of children, however their cohort had a significantly smaller range of diabetes duration than the present study. In the present study, D_LCO was lower in a cohort free of microvascular complications with relatively long diabetes duration (14.8 ± 7.3 vs. 5.0 ± 4.4 years), which is considerably longer than previous studies.

The most novel aspect of our study design was our measurement of D_LCO during peak exercise. Alveolar ventilation and pulmonary capillary perfusion are maximized during peak exercise and the demand for oxygen transfer is greatest in this state. Therefore small, often insignificant differences in resting D_LCO can increase when these systems are stressed. Studies in adults have confirmed greater impairment in pulmonary diffusion in the exercised vs. resting state that have been attributed to reduced D_M [33], V_C [17], or both [13]. However these findings are limited to adults with poor glycemic control [13, 33]. The present study showed a significant reduction in D_LCO and D_M, but not Vc during exercise in patients (young and older) with a long duration of diabetes, suggesting that diabetes affects the pulmonary capillary membrane in a manner independent of age. The lack of difference in Vc may reflect the fact that cardiac output was not lower in our diabetic cohort, as seen in other studies [13, 33]. Vracko et al. [10] and Weynand et al. [11] described thickened alveolar and capillary basal laminae in post-mortem diabetic lungs, which also occurred in the renal and muscle capillaries. It is generally believed that these permanent changes in basal membrane protein structure accumulate over time [36]. Our data in the diabetic lung support this hypothesis.

One unexpected finding of this study was a reduction in pulmonary capillary blood volume during high intensity exercise in the young subjects with diabetes. We believe the drop to partly be due to the innate variability of the measurement, but believe it could also be due to the subjects exhibiting a mild hypoxic response to the testing being performed at moderate altitude (~2100 meters). Our changes from rest to moderate exercise are of nearly identical magnitude to those described previously [20, 37]. We are not sure why we didn’t see increases from moderate to high intensity exercise, but the possibility of hypoxic vasoconstriction from exercising at moderate altitude, may be one possible explanation to why further increases in V_C at the highest exercise intensity were not seen [38].

An unusual aspect of this study is that it was conducted at moderate altitude (2100 m). The lower inspired PO₂ found at the moderate altitude impairs oxygen diffusion by reducing the alveolar-arterial O₂ gradient [39]. Chance et al. [40] have suggested that the non-symptomatic reduction in D_LCO during sea-level rest might become functionally limiting when diabetic individuals exercised at increased altitude. However, we found that our diabetic cohort had similar decreases in D_LCO at altitude to those who have performed similar studies at or near sea level [33]. Moreover, any reductions in D_LCO were not sufficient to result in a difference in S_aO₂ (data not shown) or exercise capacity between the groups. This may reflect the relatively minor hypoxia (P_A0₂ ~ 65 – 75 mm Hg) at this altitude and/or the high level of metabolic control and healthy lifestyle of our participants.

This study found that pulmonary diffusing capacity was significantly affected in people with longer diabetes duration, who had reduced D_LCO, which was exacerbated by high-intensity exercise. The lower D_LCO was associated with a decreased D_M, but not Vc, suggesting that the ‘type 1 diabetes effect’ related to changes in the alveolar capillary membrane. Despite the fact that experiments were conducted at moderate altitude, differences in pulmonary diffusion did not result in lower arterial saturation during exercise in people with long-duration diabetes. Our interpretation of these findings is that pulmonary diffusion, though impaired, does not limit aerobic capacity in people with well-controlled type 1 diabetes.