Diagnosis and Significance of Pulmonary Microvascular Disease in Diabetes

Exercise capacity, quantified as maximal oxygen consumption (VO₂peak), is a major determinant of longevity and quality of life (1). As compared with healthy individuals, VO₂peak has been observed to be reduced by ∼20% in subjects with diabetes (2,3), yet the mechanisms responsible for this functional impairment are incompletely understood. Subclinical left ventricular (LV) dysfunction (3,4), alterations in the peripheral vasculature (5), and abnormalities of the skeletal muscle microcirculation (6) have all been associated with exercise impairment in diabetes. More recently it has been recognized that the right ventricle (RV) and pulmonary circulation are placed under disproportionate hemodynamic stress during exercise and are major determinants of exercise capacity (7,8). Consideration of the lung as a “target organ” in diabetes has increased in recent years (9,10), which is not surprising given that pulmonary circulation has the largest capillary network in the body (11) and accommodates the entire cardiac output. Pulmonary microvascular pathology, including alveolar epithelial and pulmonary capillary basement membrane thickening, has been described in autopsy studies (12,13), with structural abnormalities similar to those observed in the diabetic kidney and retina (11). Despite this, it remains a challenge to directly associate pulmonary microvascular disease and cardiovascular performance.

A promising noninvasive measure of pulmonary microvascular function is the measurement of pulmonary transit of agitated contrast (PTAC) during exercise echocardiography (14). This involves the injection of an agitated colloid contrast solution and semiquantitative grading of the contrast that traverses the pulmonary vasculature to reach the left heart during exercise. We have previously demonstrated that greater PTAC (more contrast solution traversing the pulmonary circulation) is associated with higher pulmonary vascular compliance, lower pulmonary vascular resistance, improved RV function, greater cardiac output, and higher VO₂peak (14). We therefore concluded that the PTAC phenomenon may be a marker of healthy pulmonary microvascular distensibility. This hypothesis was supported by Lalande et al. (15), who demonstrated an association between PTAC and the distensibility coefficient of the pulmonary circulation. Thus, lesser transit of contrast through the lungs is associated with stiffer pulmonary microcirculation.

We hypothesized that PTAC would be reduced in people with diabetes, particularly those with evidence of microvascular disease in other vascular territories. We also sought to assess the impact of PTAC on RV function, cardiac output, and exercise capacity.

Subjects with type 1 and type 2 diabetes attending specialist hospital diabetes outpatient clinics were invited to participate in the study via promotional leaflets and approval by their treating doctor. Inclusion criteria were age 18–70 years and the ability to perform moderate-intensity exercise. Half of the subjects with type 1 and type 2 diabetes were required to have a diagnosis of microvascular complications, defined by retinopathy and/or nephropathy according to current guidelines (16).

Age- and sex-matched control subjects without diabetes were recruited in a 2:1 ratio, half matched to participants with type 1 diabetes and the other half to type 2 diabetes. For both subjects with diabetes and control subjects, exclusion criteria included chronic obstructive airway disease, coronary artery disease, LV systolic dysfunction (defined as LV ejection fraction [LVEF] <40%), or significant nephropathy (estimated glomerular filtration rate [eGFR] <30 mL/min/1.73 m²). Subjects were further excluded after completing the exercise studies if echocardiographic images were nondiagnostic.

After enrollment, a detailed baseline assessment of microvascular, cardiovascular, and pulmonary disease was performed in all subjects, including serum creatinine, urinalysis, retinal imaging for grading of diabetic retinopathy, 24-h blood pressure monitoring, echocardiography, and respiratory function testing. Participants estimated their average weekly exercise participation in the month leading into the study and the intensity of exercise (mild, moderate, or high) to calculate metabolic equivalent (MET) hours.

The study was approved by the human research ethics committee at St Vincent’s Hospital Melbourne. Written informed consent was provided by all participants.

Study participants performed two tests over a 1-week period: a cardiopulmonary exercise test (VO₂peak) and semisupine bicycle exercise echocardiography study. At least 24-h recovery was mandated between the tests.

VO₂peak testing was performed on an upright bicycle (Excalibur Sport; Lode, Groningen, the Netherlands) using an individualized continuous incremental ramp protocol with an aim of reaching peak power within 10 min. Breath-by-breath analysis of oxygen consumption and carbon dioxide production (JLab; CareFusion, Höchberg, Germany) was averaged over five breaths with simultaneous electrocardiogram (Norav Medical, Delray Beach, FL) and blood pressure (Tango M2; SunTech Medical) measures. Respiratory exchange ratio (RER), ventilatory threshold, and ventilatory efficiency (minute ventilation/volume of carbon dioxide [V_E/VCO₂]) were calculated by standard measures (17).

Exercise echocardiography was performed on a semisupine bicycle ergometer with lateral tilt (Lode). A series of images were acquired at rest and during four exercise stages at increasing power as determined from an individual’s VO₂peak. We have previously demonstrated that 66% of maximal power obtained on an upright ergometer consistently corresponds to near-maximal capacity on a semisupine ergometer (18), and thus low-, moderate-, high-, and maximal-intensity exercise was prescribed as 15%, 25%, 50%, and 66% of VO₂peak, respectively. An agitated contrast solution of succinylated gelatin (Gelofusine; Braun) mixed with room air (95:5 ratio) was vigorously agitated between two syringes using a three-way tap, and a 2 mL-bolus was injected over ∼1 s via an upper-arm venous cannula. Two boluses were administered in each stage of exercise: the first to enhance the Doppler signal for measurement of peak tricuspid regurgitation velocity, and thus pulmonary artery systolic pressure (PASP), and the second for PTAC quantification. PTAC was assessed over ∼20 beats after the first appearance of contrast in the right heart. In subjects identified to have a patent foramen ovale (PFO) during baseline testing or exercise, PTAC measurement was not included in the final analysis; a PFO was diagnosed by the appearance of contrast in the left heart in less than six cardiac cycles. Echocardiography was performed by a single experienced cardiac sonographer using a Vivid E9 cardiac ultrasound machine (GE Healthcare). A single cardiologist with expertise in echocardiography analyzed all images offline using EchoPAC software (version 113; GE Healthcare). PTAC was semiquantitatively graded into one of four grades as described previously (14).

Data were analyzed using IBM SPSS version 22 (SPSS Inc., Chicago, IL). Analysis of data normality was determined by the Shapiro-Wilk test. Continuous data were expressed as the mean value ± SD. Differences between groups were assessed using unpaired Student t tests or χ² Fisher exact test. Spearman correlation evaluated univariate associations, and a forward step binomial regression model was used to identify significant predictors of PTAC. Two-way repeated-measures ANOVA was used to assess the effect of diabetes and exercise stage on echocardiographic measures, with measurements expressed as mean value ± SE. A P value <0.05 was considered statistically significant. Based on previous observations with PTAC, we predicted that the proportion of subjects with high PTAC would be reduced by half in the group with diabetes and similarly reduced in those with microvascular disease. A sample size of n = 20 for each respective group gave 86% power to detect a difference in the proportion with high PTAC.

Sixty-four subjects were enrolled in the study, with three excluded due to suboptimal echocardiographic image quality (two with diabetes and one control subject), and a further participant withdrew prior to completion of the exercise protocols. A total of 40 participants with diabetes (20 with type 1 and 20 with type 2 diabetes, half with microvascular complications in each subtype) and 20 age- and sex-matched control subjects were included in the final analysis (Table 1). Subjects with diabetes were middle-aged (44 ± 13 years), predominantly male, and overweight. Glycemic control was moderate, and the average duration of diabetes was 16 ± 10 years. Compared with subjects with diabetes without microvascular complications, those with complications had greater BMI (29 ± 4 vs. 26 ± 3 kg/m², P = 0.038), worse glycemic control (HbA_1c 66 ± 18 vs. 56 ± 8 mmol/mol, P = 0.022), and longer duration of diabetes (21 ± 8 vs. 10 ± 7 years, P < 0.0005). The 20 subjects with diabetes with microvascular complications had retinopathy (n = 9, 45%); microalbuminuria (n = 5, 25%); retinopathy and microalbuminuria (n = 3, 15%); and retinopathy, microalbuminuria, and neuropathy (n = 3, 15%). There was no significant difference in weekly exercise participation according to estimated MET hours between groups.

LV and RV systolic function was normal in subjects with diabetes and control subjects (Table 1). Resting hemodynamic indices, including cardiac index (CI) and PASP, were within normal limits and similar between groups, although PASP was higher in subjects with diabetes with complications compared with those without (29 ± 6 vs. 24 ± 6 mmHg, P = 0.022). Five subjects were noted to have a PFO (one with diabetes and four control subjects).

Forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) were lower in subjects with diabetes compared with control subjects, whereas FEV1/FVC was similar (Table 1). Contrary to expectations, there was no difference in diffusing capacity of the lungs of carbon monoxide (DLCO) with or without correction for alveolar volume (DLCO/VA).

According to microvascular complication status, those with complications had a paradoxically higher but clinically insignificant DLCO/VA (5.0 vs. 4.5 mL/mmHg/min/L, P = 0.023) but no difference in other respiratory function measures.

All subjects exercised to high intensity during VO₂peak testing, supported by an RER >1.1 (Table 1). Maximum power and VO₂peak were significantly lower in subjects with diabetes compared with control subjects. Ventilatory efficiency was normal in both groups, although tending to be less efficient in diabetes (V_E/VCO₂ = 24.3 ± 3.3 vs. 22.5 ± 3.8, P = 0.06).

Appropriate augmentation of biventricular systolic function and hemodynamic indices was observed during incremental exercise in both groups (Supplementary Table 1). Exercise-induced increases in PASP were similar between the group with diabetes and the control group. When compared with control subjects, augmentation of the CI was less in subjects with diabetes (interaction, P = 0.003), due to a lower stroke volume throughout exercise (main effect, P = 0.007) and an attenuated increase in heart rate (interaction, P = 0.01). The increase in LV septal peak systolic velocity (LVs′) was slightly diminished during exercise in subjects with diabetes as compared with control subjects, whereas all other functional measures were similar (Supplementary Table 1).

PTAC was measured in 55 subjects. Low PTAC was recorded in 18 subjects (33%, 16 with diabetes and 2 control subjects), whereas the remaining 37 (67%) demonstrated high PTAC (23 with diabetes and 14 control subjects) (Table 2). Subjects with low PTAC were older and heavier compared with those with high PTAC. A significantly greater proportion of subjects with diabetes (16 of 39, 41%) had low PTAC compared with the control group (2 of 16, 12%; χ² P = 0.041), and low PTAC was observed more frequently in subjects with diabetes with microvascular complications (11 of 20, 55%) compared with those without complications (5 of 19, 26%) or control subjects (2 of 16, 13%; χ² P = 0.02) (Fig. 1).

Several resting echocardiography measures of LV and RV systolic function were slightly reduced in subjects with low PTAC relative to subjects with high PTAC. Resting measures of CI and PASP were similar (Table 2). As compared with subjects with high PTAC, DLCO was reduced in the subjects with low PTAC, and FEV1 tended to be less.

When compared with the high PTAC group, VO₂peak and maximal power (W) were 24% and 17% lower in the low PTAC group, respectively (P < 0.05) (Fig. 2 and Table 2).

Low PTAC was associated with higher PASP during exercise when compared with those with high PTAC (60 ± 10 vs. 52 ± 8 mmHg; interaction with exercise P = 0.002) (Table 2), whereas echocardiographic measures of RV systolic function were reduced in those with low PTAC (RV free wall peak systolic velocity [RVs′] main effect, P < 0.0005; RV fractional area change [RV FAC] main effect, P = 0.015). LVs′, LVEF, and CI were not significantly different throughout exercise (interaction, P = 0.38, P = 0.49, and P = 0.42, respectively).

Demographic markers, including age (r = −0.52, P < 0.0001) and BMI (r = −0.39, P = 0.003), were significant univariate correlates of PTAC, whereas sex was not (r = 0.19, P = 0.16). Among subjects with diabetes, current control (HbA_1c) and diabetes duration were not significant univariate correlates of PTAC. The presence of microvascular renal dysfunction was moderately correlated with PTAC (eGFR r = 0.48, P = 0.002; microalbuminuria r = −0.40, P = 0.011), whereas there was no correlation with the presence of retinopathy.

Exercise capacity (VO₂peak) was moderately correlated with PTAC (r = 0.40, P = 0.002), as were resting LV and RV systolic function measures (LVs′ r = 0.40, P = 0.002; LV global longitudinal strain [GLS] r = −0.36, P = 0.008; LVEF r = 0.28, P = 0.04; RVs′ r = 0.42, P = 0.002). Resting cardiac output and PASP had no significant correlation with PTAC. Similarly, neither DLCO (r = 0.18, P = 0.20) nor any other respiratory function parameters were associated with PTAC.

Binomial regression was performed to assess the significance of diabetes on predicting low PTAC. After adjustment for age, the presence of diabetes remained a significant covariate. The combined model explained 46% of the variance (P < 0.0001) and correctly classified 78% of cases.

We used a novel surrogate of pulmonary microvascular function to demonstrate more prevalent impairment in pulmonary vascular and RV function among subjects with diabetes than among control subjects. This was associated with older age, lower exercise capacity, and lower VO₂peak. This suggests a novel mechanism of pulmonary microvascular pathophysiology and resulting cardiovascular limitation in subjects with diabetes.

The pulmonary vascular bed must swiftly accommodate significant increases in blood volume and thus cardiac output during exercise. Studies of dog lungs (19,20) have suggested that there is recruitment of new capillary beds and distention of existing capillaries, with an increase in mean capillary diameter of 2% for every mmHg rise in vascular pressure (21,22). This additional recruitment and distensibility of the pulmonary microcirculation attenuates the rise in mean pulmonary artery pressure during exercise, resulting in a curvilinear relationship between mean pulmonary artery pressure and cardiac output (23–25). Malhotra et al. (26) observed a strong relationship between pulmonary vascular distensibility, RV function, exercise capacity, and survival in patients with heart failure or pulmonary hypertension. They derived pulmonary vascular distensibility by combining invasive pulmonary artery and capillary wedge pressure measurements with radionuclide ventriculography estimates of cardiac output during exercise. These findings, although intriguing, are limited in clinical application due to the invasiveness and complexity of the measures. In our current study, we also identified an association among pulmonary vascular distensibility, RV function, and exercise capacity but did so using a simple methodology that has the potential to be incorporated into routine clinical care with relative ease and minimal cost.

The study of pulmonary microvascular dysfunction seems particularly pertinent in diabetes given that the functional consequence of pulmonary microangiopathy remains largely unknown. Reductions in DLCO have been inferred to represent pulmonary microangiopathy, although there have been conflicting results regarding the sensitivity of DLCO and respiratory function testing at rest to identify early alveolar-capillary injury (10,11,27). To some extent, our current finding of reduced DLCO in subjects with low PTAC is consistent with the argument that both measures represent pulmonary microvascular dysfunction. On the other hand, we did not observe abnormalities in DLCO in the whole diabetes cohort when compared with healthy control subjects and yet there was reduced PTAC in subjects with diabetes. The lack of agreement is difficult to explain. It is possible that PTAC is a more sensitive measure than DLCO and that our cohort with relatively preserved functional capacity (VO₂peak = 93 ± 20% of predicted) and generally mild microvascular disease may have had only mild pulmonary microangiopathy. Alternatively, measurement of DLCO during exercise as suggested by Wang et al. (28) may have offered a better reflection of the true gas diffusion capacity of the lung to parallel the differences observed in PTAC. The hypothesis that PTAC could identify pulmonary microvascular disease at the very earliest stages is intriguing but there is no easy means of validating this given that there is no gold standard diagnostic technique against which to reference.

Our most significant finding is the improvement in exercise capacity, both power and VO₂peak, that was associated with high PTAC. The rationale for undertaking the study was based on our previous findings that low PTAC was associated with reduced exercise capacity and that all noninvasive indicators suggested that this association was mediated through an increase in pulmonary vascular resistance and a reduction in compliance (14). The hypothesis that this would be particularly pertinent among subjects with diabetes in whom we could anticipate pulmonary microvascular disease proved correct. Historically, there has been a disproportionate focus on LV determinants of exercise capacity. However, there have been few associations found between LV functional measures and exercise capacity, and failed attempts at validation have left little consensus on what cardiac features define “diabetic cardiomyopathy.” This diagnosis is made by exclusion, when “ventricular dysfunction occurs in the absence of coronary atherosclerosis and hypertension” (29), highlighting the lack of diagnostic markers on cardiac imaging. This also raises the possibility that the cardiovascular limitation may reside outside of those considered with conventional measures. It has been established that the load on the right-sided heart chambers increases disproportionately during exercise (30,31), and it has been argued that the RV represents an “Achilles’ heel” in the cardiovascular system during exercise (7,32). This appears true in the healthy circulation and is exacerbated in the setting of pulmonary vascular dysfunction. In our current study, these physiological principles are supported by the finding that low PTAC was more prevalent and pulmonary artery pressures were higher in subjects with diabetes, suggesting pulmonary microvascular dysfunction and a resultant increase in RV afterload, respectively. These conditions were associated with reduced RV functional measures and the consequence of impaired exercise capacity. This schema provides an entirely novel explanation for exercise intolerance in people with diabetes (Fig. 3).

The unifying physiological mechanism of PTAC remains speculative. Some have proposed that the presence of PTAC reflects intrapulmonary shunting through arteriovenous pathways (33–35). However, arteriovenous malformations have not previously been associated with diabetes, whereas microvascular disease is a pathognomic manifestation. Thus, the association between low PTAC and diabetes strengthens our hypothesis that PTAC reflects pulmonary capillary distensibility. Microvascular disease rarely occurs in isolation in diabetes, and the greater prevalence of low PTAC in those with microvascular disease in other vascular territories further supports the argument that low PTAC is a marker of pulmonary microvascular pathology.

We observed a significant association between low PTAC and increasing age, consistent with the findings of other investigators (14,15,24). Harris et al. (36) and Mackay et al. (37) have previously reported a progressive increase in pulmonary arterial wall stiffness with increasing age, and Mackay et al. propose that this is due to alterations in elastic tissue over time. However, age alone does not explain all of the reduction in pulmonary microvascular compliance. When matched for age with control subjects, there was a significantly higher prevalence of low PTAC among subjects with diabetes. Furthermore, our regression model identified both age and diabetes as independent predictors of PTAC. It can therefore be concluded that diabetes exacerbates age-related impairment in pulmonary capillary compliance, thereby contributing to reduced exercise capacity.

The associations among low PTAC, diabetes, microvascular complications, and lesser exercise capacity suggest that this simple method may be a sensitive marker of microvascular dysfunction with prognostic and functional implications. Furthermore, this suggests an entirely novel mechanism of exercise limitation in people with diabetes in which the pulmonary circulation and RV may be a dominant source of limitation. The potential efficacy of targeted anti-inflammatory and antifibrotic agents could be measured using PTAC (38). It has also been suggested that inhaled insulin may cause inflammation and injury of the lungs and pulmonary microvasculature (39,40), which could be monitored using PTAC rather than traditional lung function testing. Our data suggest that impaired exercise capacity would be a consequence and that PTAC would be a sensitive means of identifying early pathology.

Like other studies assessing the pulmonary microvasculature noninvasively, we are limited to inferences in the mechanism of PTAC. We did not perform arterial sampling to assess effects of PTAC on the alveolar/arterial oxygen diffusion gradient, although we have previously shown that PTAC does not influence exercise-induced changes in arterial oxygen saturation in healthy subjects (14).

Cardiopulmonary and semisupine bicycle echocardiography exercise tests were performed separately, which may have resulted in subtle differences in peak exercise performance between tests. However, in our experience, the combination of tests is technically more challenging, resulting in compromised image acquisition and exercise performance. The brief time interval between exercise tests minimizes the likelihood of having significant differences in peak exercise performance.

It is difficult to completely exclude confounders of flow and pressure on the survival and destruction of the microbubbles. However, in a previous study, we found no difference in PTAC according to invasive assessments of LV afterload (14).

The sample size and heterogeneity of the subjects may have limited power, although our study represents the largest exercise study to include comprehensive cardiac and pulmonary vascular measures in subjects with diabetes. Both the control cohort and cohort with diabetes had well-preserved functional capacity, and this may have been expected to reduce the differences in PTAC and cardiovascular measures. Furthermore, subjects with type 1 and type 2 diabetes were combined in the analysis, as microvascular disease does not discriminate between the two subtypes. As is typical of studies including exercise assessment, it is possible that fitter subjects volunteered to participate. The strength of association between low PTAC and microvascular pathology may have been stronger if a larger population of patients with more severe microvascular disease had been studied. It is notable that the significant association between PTAC and cardiovascular function was observed in an outpatient population with only mild diabetic complications.

We have demonstrated that subjects with diabetes have less PTAC, particularly in the presence of microvascular complications, and that this is associated with reduced RV function, higher pulmonary artery pressures, and lower exercise capacity. Thus, PTAC may be a simple surrogate measure of pulmonary microvascular disease with important implications for cardiovascular function and exercise capacity.

Acknowledgments. The authors acknowledge the Statistical Consulting Centre at The University of Melbourne for overseeing the statistical methodology chosen to analyze the data presented in this manuscript, D.J. Mooney (St Vincent’s Hospital Melbourne) for performing all echocardiography studies, N. Cohen (Baker Heart and Diabetes Institute), A. Jenkins (The University of Sydney), D. O’Neal (The University of Melbourne), G. Ward (The University of Melbourne) for subject recruitment, D.J. Campbell (St Vincent’s Institute of Medical Research) for study design support, and J. Barros (The University of Melbourne) for assistance during exercise studies.

Funding. This study was funded by the National Heart Foundation of Australia (Grant-In-Aid G 12M 6396). A.L.G. is supported by a Career Development Fellowship from the National Health and Medical Research Council (NHMRC 1089039) and a Future Leaders Fellowship from the National Heart Foundation of Australia (NHF 100409).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. T.J.R. designed the study, recruited subjects, acquired and analyzed data, and wrote the manuscript. A.T.B., A.I.M., and D.L.P. edited the manuscript. R.J.M. recruited subjects and edited the manuscript. A.L.G. conceived and designed the study, recruited subjects, analyzed data, and wrote and edited the manuscript. T.J.R. and A.L.G. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.