Vascular volume changes in radiological patterns of usual interstitial pneumonia in patients with type 2 diabetes

Abstract

Objective

This research primarily focuses on exploring the changes in intrapulmonary vascular volume (IPVV) in radiological patterns of usual interstitial pneumonia (UIP) associated with Type 2 Diabetes Mellitus (T2DM), thereby inferring the possible mechanisms of the co-occurrence of diabetes and UIP patterns.

Methods

Thin-layer data were post-processed on the basis of high-resolution computed tomography (HRCT) and quantitatively assessed for IPVV. Changes in IPVV were compared between T2DM combined with UIP modality and T2DM non-UIP modality. Correlations between UIP patterns and various markers and confounders, including IPVV, were determined via logistic regression analysis. In this study, the potential of IPVV as a predictor for UIP presence was analysed through the application of subject operating characteristic curve analysis.

Results

In patients with T2DM, the IPVV demonstrated smaller size in those with combined UIP patterns compared to T2DM patients without UIP patterns (164.4 ± 68.7 vs 202.9 ± 76.3 mL, P = 0.005). We detected a positive correlation between IPVV levels and several variables, including fasting plasma glucose (FPG) (r = 0.404, P < 0.0001), glycated hemoglobin (HbA1c) (r = 0.225, P = 0.022), serum uric acid (SUA) (r = 0.332, P = 0.0007) and HRCT scores (r = 0.288, P = 0.024). Conversely, negative correlations were noted with total cholesterol (TC) (r = –0.220, P = 0.028) and cystatin-C (Cys-C) (r = –0.215, P = 0.038). Multivariate logistic regression analysis identified independent associations between the presence of UIP and several factors: IPVV, age, smoking history, and FPG. In assessing the combined UIP pattern among T2DM patients, IPVV levels exhibited a sensitivity of 70.5% and a specificity of 58.5%, generating an AUC of 0.645.

Conclusion

In individuals diagnosed with T2DM alongside UIP, a substantial decline in IPVV was documented. This diminution correlates with the presence of UIP, suggesting that IPVV may serve as a potent biomarker for detecting UIP patterns in individuals with T2DM. This may suggest that the mechanism behind the co-occurrence of T2DM with UIP patterns is attributed to alterations in the pulmonary microvasculature, potentially representing one of the vascular complications associated with diabetes.

Introduction

Diabetes mellitus (DM) represents a metabolic disorder characterized by persistent hyperglycemia, attributed to defective insulin secretion or action. Type 2 diabetes mellitus (T2DM) predominantly affects adults, which is the most prevalent form of diabetes (constituting over 90% of diabetic cases). It is a complex group of diseases associated with insulin resistance and pancreatic β-cell dysfunction [1], with complications that often involve multiple organs and systems, resulting in poor prognoses. Lungs are rich in connective tissue and alveolar capillary networks, and they may become the target of microvascular injury in diabetes. In the 1970s, foreign scholars first suggested that the lungs may be attacked by diabetes [2], thereby drawing attention to diabetic lung injury. The poor prognosis of pulmonary fibrosis in diabetic lung injury necessitates attention. Its mechanism may involve hyperglycemic conditions activating multiple pathways, leading to intracellular stress and abnormal cytokine expression. This results in the loss of basal layer integrity in the alveolar-capillary barrier, failure of re-epithelialization and re-endothelialization, and epithelial-mesenchymal transition, ultimately causing pulmonary structural destruction and pathological fibrosis [3,4,5,6].In 2021, Kopf et al. documented evidence indicating that pulmonary fibrosis may occur as a complication associated with diabetes, more pronounced in T2DM, and strongly recommended a systematic study of pulmonary fibrosis in diabetic patients, especially in the context of the radiological pattern of usual interstitial pneumonia (UIP) [7]. Therefore, pulmonary fibrosis is not only a significant comorbidity of T2DM but also a potential vascular complication. The mechanisms and imaging studies of this condition are key areas of interest. This study focuses on T2DM-associated pulmonary fibrosis, aiming to verify whether pulmonary vascular injury in T2DM patients could be a possible mechanism underlying the presence of pulmonary fibrosis.

The computed tomography (CT) has high-resolution capabilities to visualize lung structures in detail and quantify various lung abnormalities. Several studies have been carried out to quantitatively assess pulmonary vascular and airway changes in chronic obstructive pulmonary disease (COPD) and emphysema on the basis of high-resolution computed tomography (HRCT) [8, 9]. Furthermore, previous studies [10,11,12] have shown that the quantitative evaluation of pulmonary vascular changes adopts methods including the automatic extraction of vascular trees and the segmentation calculation of intrapulmonary vascular volume (IPVV) and total cross-sectional area (CSA) of small pulmonary vessels. Although Jacob et al. [12] confirmed the correlation of idiopathic pulmonary fibrosis (IPF) disease severity using a quantitative analysis of IPVV, it is not clear that IPVV correlates with T2DM pulmonary fibrosis disease.

This study was designed to quantify IPVV in T2DM patients on the basis of HRCT. The aim of our research was to examine changes in IPVV and the factors influencing UIP in patients with T2DM. Additionally, this research aimed to ascertain whether IPVV could serve as a predictive biomarker for T2DM-UIP. This is the first clinical study to elucidate the changes in IPVV in T2DM and analyse the relationship between IPVV and T2DM-UIP by conducting a quantitative assessment of the pulmonary vasculature.

Methods and materials

Human subjects

From February 2019 to August 2023, data were collected at the Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, from 61 patients who had been clinically and radiologically diagnosed with T2DM-UIP, and from 53 patients who had been clinically diagnosed with T2DM but exhibited no radiological evidence of abnormal lung changes. The inclusion criteria for the study were defined as the following: (1) a diagnosis of type 2 diabetes mellitus (T2DM) under the 2019 guidelines set by the World Health Organization (WHO) [13]; (2) patients aged ≥ 18 years; (3) HRCT was performed on all patients. Patients with metabolic diseases other than obesity, T2DM and severe UIP were excluded. Patients were excluded from this study due to other conditions that cause interstitial lung disease (ILD) and IPF, such as connective tissue disease (CTD), COPD, lung infections, tuberculosis and lung tumours. 61 healthy controls were from the hospital's healthcare staff who were of similar age and gender distribution as the patients in the T2DM group. All data were accessed through the Hospital Information System (HIS) and the Picture Archiving and Communication System (PACS). Ethical approval for our research was obtained from the Ethics Committee at the Shanghai Sixth People’s Hospital, affiliated with Shanghai Jiao Tong University School of Medicine [approval number: 2024-KY-017(K)].

Patient clinical assessment

Collect comprehensive demographic characteristics, laboratory test results, current treatments, and clinical features of all patients. These include age, sex, disease duration, body mass index (BMI), glucose-related indicators, lipid-related markers, renal function parameters, smoking history, and history of insulin therapy.

HRCT scan

All HRCT scans were conducted using Siemens' third-generation dual-source CT scanners (SOMATOM Force; Siemens AG, Germany). These devices proficiently reconstructed 1 mm HRCT images and efficiently transmitted the data to the PACS workstation.

Clinical assessment of patients and UIP diagnosis

Diagnosis of UIP radiological pattern based on HRCT. In 2018, the Fleischner Association provided the latest HRCT diagnostic criteria [14], which included honeycomb opacities, traction bronchiectasis, reticular opacities, ground glass opacities, and non-emphysematous cysts, mainly in the basal and subpleural regions. Overall CT scores were derived by summing the six mean scores, which were assigned by the radiologist in three regions of each lung. The HRCT scan protocol is delineated by three key axial levels: the upper boundary of the aortic arch, the location of the tracheal carina, and a point 1 cm superior to the diaphragm [15]. Quantify and assess the extent of fibrosis across six layers within the pertinent lung field, relative to the lesion's affected area [16] (Table 1). Two radiologists with six and 10 years of experience in chest radiology added up the scores of the six bilateral regions to obtain the overall CT score, with the final score determined by consensus.

Table 1 Lung tissue involvement criteria

Image post-processing

IPVV was extracted using 3D Slicer segmentation software via the following processes: (1) input DICOM image, select 1 mm image and select lung window (–600/1600HU).; (2) threshold segmentation, which involves selecting pixels as –700–734HU; (3) separate the chest and major cardiovascular and pulmonary vessels layer by layer and optimise the separation of pulmonary blood vessels (Fig. 1); (4) output a three-dimensional image and measure the volume in millilitres and (5) take the average of three consecutive measurements of IPVV.

Fig. 1

Cluster separation of intrapulmonary vessels. A Identification of the thorax and cardiothorax; B heart and cardiothoracic vessels; C intrapulmonary vessels

Statistical analysis

The data were statistically investigated using PRISM (version 9) (GraphPad Software, La Jolla, CA, USA) and SPSS (version 27.0) (SPSS Inc., Chicago, IL, USA). The error bars in the bar chart were represented by the standard error of the data. Differences between groups were analyzed by Student’s t-test. Categorical variables were compared using the Pearson chi-square test. In the case of non-parametric data, results were presented as medians (minimum–maximum), and the Mann–Whitney U test was used to examine the group differences. The correlation between the two groups was identified using Spearman’s correlation coefficient. To ascertain factors associated with UIP presence, we conducted a univariate logistic regression analysis. Multivariate logistic regression analysis was performed; this analysis incorporated confounders that demonstrated significant associations in the univariate analysis. The factors were gradually selected, because of the limited number of events in the logistic model. To establish the validity and optimal cutoff value of a variable, a receiver operating characteristic (ROC) curve was employed. Data were reported as the mean ± standard deviation. P-values below 0.05 were deemed to indicate statistical significance.

Results

Clinical and demographic profiles

This research recruited 114 T2DM patients (61 with concomitant UIP and 53 without UIP) and 61 age- and gender-matched healthy controls (HC). The median course of the patients with T2DM was 10.5 years, varying from 1 to 40 years. The age average was 69.2 ± 8.4 years. To explore the alterations in intrapulmonary vascular volume (IPVV) among T2DM patients and concurrent usual interstitial pneumonia (UIP), participants were categorised into two cohorts: those with T2DM and UIP (T2DM-UIP, n = 61) and the control group comprising T2DM patients without UIP (T2DM-non-UIP, n = 53). Table 2 presents a comparative analysis of the data collected for T2DM-UIP and T2DM-non-UIP patients. Meaningful distinctions were noted between the T2DM groups with and without UIP in terms of age, smoking history, glycated haemoglobin (HbA1c), fasting blood glucose (FPG), baseline C-peptide levels, fasting insulin, homocysteine (Hcy) and lactate dehydrogenase (LDH).

Table 2 Characteristics of T2DM patients with and without UIP patterns

IPVV increased in T2DM-non-UIP patients, and on this basis, T2DM-UIP decreased

IPVV was significantly reduced in T2DM-UIP patients (164.4 ± 68.7 mL) compared to in T2DM-non-UIP patients (202.9 ± 76.3 mL, P = 0.005) (Fig. 2) (Fig. 3A), who showed a significant increase in IPVV compared to the HC (174.3 ± 51.3 mL, P = 0.019) (Fig. 3A). IPVV levels exhibited a moderate positive correlation with FPG (r = 0.404, P < 0.0001) (Fig. 3B). Further correlation analyses were carried out to determine the correlation coefficients between IPVV and other clinical markers. Significant correlations were found for HbA1c, total cholesterol (TC), SUA (serum uric acid), SUA/SCr (serum uric acid/serum creatinine) ratio and cystatin C (Cys-C). The correlation coefficients between IPVV and HbA1c, TC, SUA, SUA/SCr, and Cys-C were r = 0.225 (P = 0.022), r = –0.22 (P = 0.028), r = 0.332 (P = 0.0007), and r = 0.22 (P = 0.028), respectively, r = –0.215 (P = 0.038) (Fig. 3C–G). In addition, we identified a negative correlation between IPVV and HRCT scores (r = 0.288, P = 0.024) (Fig. 3H).

Fig. 2

Comparison of IPVV for T2DM-UIP and T2DM-non-UIP in HRCT. A T2DM-UIP patient (male, 74 years old): inspiratory HRCT whole-lung IPVV was 133.08 ml. B T2DM-non-UIP patient (male, 56 years old): inspiratory HRCT whole-lung IPVV was 235.96 ml

Fig. 3

IPVV levels (A) in healthy controls and T2DM patients, along with their associations with clinical markers (B–G) and HRCT scores (H)

IPVV was related to the presence of UIP in T2DM

We conducted multivariate and univariate logistic analyses to identify factors linked to UIP in T2DM patients. According to univariate analysis, IPVV, age, smoking history, FPG, HbA1c, baseline C-peptide levels, Hcy, Cys-C, and LDH were correlated with the presence of UIP. In multivariate analysis, after gradually choosing factors from the multivariate model (backward selection), IPVV, age, smoking history, and FPG remained independent factors associated with UIP (Table 3). The missing independent variables in the case were supplemented using the completion method, and baseline C-peptide levels and Hcy were not included in the regression model due to excessive deletions.

Table 3 Factors associated with UIP patterns in T2DM patients – logistic regression analysis

IPVV levels could serve as disease markers for detecting T2DM combined with UIP patterns

To evaluate the clinical significance of IPVV, sensitivity and specificity analyses were conducted in patients diagnosed with T2DM-UIP (Fig. 4). The area under the curve (AUC) for IPVV was 0.645 (SE: 0.052; range: 0.544–0.746; sensitivity: 0.705; specificity: 0.585). T2DM-UIP was divided into an IPVV-positive group and an IPVV-negative group according to critical values. The comparison of collected data between these groups is described in Table 4. Significant differences were observed in lung HRCT scores across the groups. Consequently, these findings indicate that IPVV may serve as an independent biomarker for detecting the presence of T2DM-UIP. Furthermore, IPVV has the potential to be a standalone biomarker for assessing disease severity and monitoring UIP progression in this patient population.

Fig. 4

ROC curve analysis for IPVV levels in terms of detecting T2DM-UIP

Table 4 Characteristics of T2DM-UIP patients by IPVV status

Discussion

This is the first report on the relationship between IPVV and the presence of T2DM-UIP. IPVV was considerably decreased in T2DM-UIP patients contrasted to T2DM-non-UIP patients. In addition, IPVV, age, smoking history and FPG were independent disease markers for detecting UIP patterns in T2DM patients.

In this study, we post-processed patient imaging data, extracted intrapulmonary vasculature, quantitatively assessed IPVV and determined a strong correlation between IPVV and UIP patterns in T2DM patients. The research demonstrated a substantial increase in IPVV in patients with T2DM-non-UIP patterns and a significant decrease in IPVV in patients with T2DM-UIP. These findings are consistent with the previously reported thickening of the pulmonary capillary basal layer in diabetes, and the mechanism of T2DM with fibrosis involves Compromised vascular re-endothelialisation and diminished integrity of the basal layer of the alveolar capillary barrier [17, 18]. Interestingly, our findings establish a notable positive correlation between elevated serum IPVV levels and HRCT scores in patients with T2DM-UIP. Multivariate logistic analysis identified reduced IPVV as a clinically significant risk factor for T2DM-UIP. The ROC curve further demonstrated the ability of IPVV as a promising biomarker in T2DM-UIP. Therefore, this paper suggests that IPVV could serve as a useful biomarker for detecting UIP patterns in T2DM patients.

Visual quantitative scoring is frequently employed in the assessment of various pulmonary diseases and is considered an indicator of disease progression. Although the HRCT score in patients with T2DM-UIP is not the gold standard for evaluating disease progression, it can indirectly assess disease severity. Our study demonstrates a positive correlation between IPVV and the HRCT score, indicating that IPVV is not only associated with T2DM combined with UIP but also related to disease progression. Therefore, IPVV may be a biomarker closely linked to the occurrence and development of pulmonary fibrosis in T2DM, suggesting that pulmonary fibrosis could be one of the vascular complications of T2DM.

Past research has shown that the radiological pattern of UIP is predominantly seen in older men with a history of smoking [19, 20]. Smoking can cause bronchial lumen narrowing and inflammatory responses, resulting in impaired air expulsion from alveoli, increased alveolar pressure, and eventual alveolar rupture or distension. Additionally, smoking may drive neutrophil aggregation in target cells, activating proteases that degrade related proteins, thereby increasing the risk of emphysema. This further damages lung tissue, promotes the secretion of large amounts of extracellular matrix and collagen, and heightens the risk of interstitial lung changes [21]. A number of serological markers serve as predictors of the progression of lung injury in DM. For example, FPG and HbA1c levels in DM patients are negatively correlated with lung function [22, 23]. Insulin resistance has adverse effects on lung function [24], This is consistent with the high baseline C-peptide levels and low insulin levels in our paper. In addition, although high Hcy levels have been reported to impair the endothelial function of the microvascular system in patients with T2DM [25], a notable and independent correlation exists between serum Cys-C levels and lung function [26]. LDH [27] is significantly increased in ILD patients, but these biomarkers are not associated with T2DM-UIP. Due to numerous missing samples, Hcy was not included in the multivariate logistic regression analysis. It was not certain whether Hcy, Cys-C and LDH were independent disease markers for T2DM-UIP, but they were demonstrated to be risk factors for T2DM-UIP and were associated with disease development. We found that in addition to IPVV, age, smoking history, blood glucose and level of glycaemic control, and insulin resistance were risk factors for UIP in T2DM, a finding that is consistent with previous reports.

One common adverse reaction to insulin therapy is that insulin resistance can lead to impaired lung function [28, 29]. This response raises concerns that insulin therapy may be associated with increased incidence or disease progression in T2DM-UIP and is less therapeutically effective than absolute insulin deficiency in type 1 diabetes mellitus (T1DM), resulting in a possible decrease in the use of insulin therapy for T2DM. Moreover, this study was limited by a small sample size and did not include assessments stratified by age and gender, nor did it evaluate the specific stages of the disease. These limitations may account for the lack of a definitive correlation between insulin therapy and T2DM-UIP observed in our findings, underscoring the need for further research to confirm these results. Obesity factors have been reported to correlate with pulmonary dysfunction in T2DM patients [30], In our study, although it was not associated with T2DM-UIP, total cholesterol levels were negatively correlated with IPVV. Therefore, we speculate that significant results may be obtained in large-scale clinical studies.

Previous research has demonstrated that elevated uric acid levels are implicated in the development of pulmonary fibrosis [31]. A recent study investigating the relationship between adult lung function parameters and SUA/SCr ratio showed a negative correlation between lung function indicators FVC and FEV1 and SUA/SCr ratio [32]. In our research, no significant difference was found in SUA levels and SUA/SCr ratios among the T2DM-UIP and T2DM-non-UIP groups, but there was a positive correlation with IPVV levels. This is different from previous studies and further in-depth research is needed.

UIP is the most common imaging pattern of pulmonary fibrosis, often indicative of end-stage fibrosis. The UIP radiological pattern can occur under various pathophysiological conditions, with autoimmune diseases recently gaining attention. Beyond CTD, UIP patterns are also observed in patients with interstitial pneumonia with autoimmune features. These patients exhibit certain CTD-related symptoms and signs, along with serological evidence of autoimmune disease, but do not meet the classification criteria for any defined CTD. This condition is considered an early stage of CTD, such as undifferentiated CTD. In autoimmune diseases, 80% of systemic sclerosis (SSc) patients exhibit pulmonary fibrosis or ILD [33]. While nonspecific interstitial pneumonia (NSIP) is the most common radiological pattern of lung involvement in autoimmune diseases, UIP patterns are also observed in some cases [34]. The pathogenesis of pulmonary fibrosis in SSc is associated with endothelial dysfunction and microvascular involvement [35], similar to the mechanisms underlying UIP patterns in T2DM. However, SSc is characterized not only by microvascular injury but also by large vessel involvement and vascular stiffening, which are widely recognized [36]. Consequently, pulmonary arterial hypertension (PAH), commonly associated with pulmonary fibrosis, is frequently observed in SSc patients. This link between PAH and pulmonary fibrosis provides insights into the reduced IPVV in patients with T2DM and UIP patterns, which may be attributed to inflammatory phenotypes involving the pulmonary vascular bed [37].

Moreover, our current study has several limitations. Firstly, biased analyses may stem from the small sample size of this study; thus, large sample of people should be studied. Second, in order to prevent the obscuring of normal vascular patterns in individuals afflicted with severe UIP, our study exclusively focused on radiologically non-severe UIP patients. However, this deliberate selection imposed certain constraints on the generalisability of our findings regarding the predictive utility of IPVV in the context of T2DM-UIP disease. Finally, while quantitative measurement of IPVV during the clinical assessment of T2DM-UIP is inconvenient and imprecise, the use of IPVV as an independent biomarker for the presence of the disease suggests that T2DM-UIP may be vascularly related.

In conclusion, our study reveals that alterations in pulmonary microvasculature constitute a crucial mechanism in the pathogenesis of T2DM concomitant with UIP patterns, extending beyond certain clinical factors. Furthermore, our findings suggest that specific clinical factors and pulmonary microvascular changes potentially influence each other, highlighting a complex interplay central to this comorbidity. To validate this conclusion, further attention and in-depth research into the mechanisms of pulmonary fibrosis in diabetes are required. It would be worthwhile to conduct disease monitoring through a quantitative evaluation of IPVV changes through prospective research design and establish a complete follow-up system to help draw more confident conclusions.