Introduction

Diabetes is a major risk factor of ischemic heart diseases. Severe coronary atherosclerosis is often silent in diabetic patients [1]. Consequently, it has been proposed to screen diabetic patients for infraclinic coronary atherosclerosis by calcium scoring in order to conduct functional testing only in the specific subgroup of high risk patients with severe infraclinic calcified coronary atherosclerosis [2–8]. Thus, Coronary Artery Calcium Score (CAC) measurement by Multiple Detector Computed Tomography (MDCT) is nowadays recommended to measure the intensity and the diffusion of silent coronary calcified plaques, to reclassify cardiovascular risk by several scientific societies [9,10].

The field of view upon CAC assessment includes the heart, the mediastinum, the upper part of the liver and the central part of the lungs. Thus, incidental findings of thoracic and abdominal lesions, within the window used for CAC, may be of interest when setting recommendations regarding the systematic screening of infraclinic coronary atherosclerosis in high risk diabetic patients [11–13]. Identification of incidental findings may provide a unique opportunity to detect neoplasic lesions which could be cured only at an early asymptomatic stage. Recently, the US Preventive Services Task Force recommended to conduct studies in real life which would explore the yield of incidental findings detection following a CAC assessment and the downstream consequences [12]. Thus, we conducted a retrospective systematic study of 732 consecutive asymptomatic diabetic subjects at high cardiovascular (CV) risk who underwent a CAC assessment (DISCO cohort). The aims of this study were to evaluate the prevalence of any incidental findings discovered, to identify the predicting factors and finally to report the midterm consequences of these findings in medical care.

Materials and methods

Results

Clinical characteristics of the subjects

732 diabetic patients were included. Patient characteristics are summarized in Table 1. They were middle aged, with a balanced sex ratio and a mild obesity. They had a long duration of diabetes: 40.2% had more than 20 years of diabetes duration. Most of the patients had TD2. They had and on average 2.1 cardiovascular risk factors and poor glycemic control. More than 40% had at least one microvascular complication. The median CAC was 29 AU [IQR 227]. 65% of the patients had a CAC value below 100. The distribution of type of diabetes among CAC groups (< 100, 100–400, > 400) was not statistically different (p = 0.510) (S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Clinical characteristics of the patients.

https://doi.org/10.1371/journal.pone.0251693.t001

Incidental findings

The breakdown of incidental findings is shown in the Fig 1. 16.0% of the diabetic patients had incidental findings including 1.3% related to the patient past history. Incidental findings were more frequent in type 3 diabetes (n = 15/63, 23.8%) and in type 2 diabetes (n = 82/482, 17.0%) than in type 1 diabetes (n = 20/187, 10.7%) (p = 0.041 for type 2 vs type 1, p = 0.009 for type 3 vs type 1, p = 0.185 for type 2 vs type 3). Only 12 incidental findings were not taken into account by the diabetologist in charge of the patient without any clear reason identified in medical reports. Eighty four patients with newly found incidental findings (77.8%) underwent additional investigations coordinated by the diabetologist in charge of the patient. The percentage of additional tomodensitometry (TDM) and specialized medical advices following the incidental findings discovery are summarized in the Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Breakdown of incidental findings.

MDCT-CAC: Multiple Detector Computed Tomography—Coronary Artery Calcium Score.

https://doi.org/10.1371/journal.pone.0251693.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Tomodensitometry and specialized advices following incidental findings discovery upon CAC assessement.

Results are presented over the whole cohort and all incidental findings (A) and splitted by type of incidental findings (B). *BPP: Bronchopulmonary and pleural, † no TDM realized but 6 cardiac echographies.

https://doi.org/10.1371/journal.pone.0251693.g002

Lung incidental findings.

Most of the incidental findings were pulmonary (70.4%). 10.4% (76/732) of these asymptomatic diabetic patients presented a lung incidental finding: 31/76 had nodules, 18 hilar lymphadenopathies, 8 emphysema, 7 pleural diseases, 4 bronchiectasis, 4 bronchitis and 4 pneumonia.

Among the 31 patients with lung nodules, 38.7% had micronodules with a size less than 5 mm, 35.5% had nodules 5–10 mm, 19.4% had nodules 10–30 mm and 2 patients had a nodule size above 30 mm. The diagnosis and the therapeutic management of patients with lung nodules are detailed in the S2 Table. 87% of patients had a thoracic TDM and 77% a specialized advice. None of the micronodules had significantly increased or had become cancerous, over an average 23 months follow-up. Three patients with nodules ≥ 30 mm,had pulmonary carcinoma (S1 File and S3 Table).

The medical cares regarding the other pulmonary incidental findings are summarized in the S4 Table.

Others incidental findings.

After lung incidentals findings, heart and vascular incidentals findings were the most frequent (12.0% and 8.3%, respectively). The details and the medical cares regarding heart, vascular, thymic, digestive and vertebral incindental findings are summarized in the S5 Table.

To summarize, 42.6% of patients with incidental findings had a TDM and 56.8% a specialized medical evaluation representing respectively 6.3% and 8.3% of the whole cohort (Fig 2A). For 10 patients (9.3% of incidental findings or 1.4% of all patients), the identification of incidental findings led to a specific treatment of the underlying disease.

Characteristics of diabetic patients with and without any incidental findings

No significant differences regarding sex, age, BMI and diabetes duration between diabetic patients with and without incidental findings were found (Table 2). Incidental findings were more often found in patients who smoked more than 20 pack-years (PY) (26.5% vs 16.5% for less than 20 PY and 14.2% for those who never smoked; p = 0.040). No significant difference regarding active smoking, weaned smoking for less than 3 years or more than 3 years was found. Incidental findings were more often found in patients with a type 3 diabetes (23.8% vs 10.7% for type 1 diabetes and 17.0% for type 2 diabetes, p = 0.028). Blood glucose control and need for insulin was similar between both groups of diabetic patients. CAC, hypertension prevalence and blood pressure levels were similar in both groups. However, diabetic patients with nephropathy had more incidental findings considering either microalbuminuria (21.2% vs 13%; p = 0.004) or renal failure (GFR < 60 ml/mn, 23.7% vs 15.1%; p = 0.053) whereas it was not the case for patients with retinopathy.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of subjects with or without incidental findings and pulmonary nodules.

https://doi.org/10.1371/journal.pone.0251693.t002

In multivariate analysis (following adjustment on age, sex, BMI, diabetes duration, HbA1c, and renal failure), microalbuminuria, type of diabetes (TD2 and TD3 vs TD1) and smoking (active or weaned vs never) remained independently associated with incidental findings (respectively OR 1.89(1.23–2.90), p = 0.003; OR 1.99(1.09–3.67), p = 0.026; OR 1.56 (1.00–2.42), p = 0.050).

Discussion

This study reports a high rate of incidental findings in our diabetic population since incidental findings were found in about one out of six patients, including mostly pulmonary findings (70.4%) and of which 28.7% were lung nodules. The exact prevalence of lung incidental findings in these patients is likely to be underestimated since the window used for the CAC assessment does not take in account the upper third of the chest.

Other studies in the general population have considered the prevalence of incidental findings found upon CAC assessment. They report a highly variable prevalence, ranging from about 8% to 25% [13,14,16,17] and even over 50% [12,18,19]. Few early studies involved Electron Beam Tomography (EBT) instead of MDCT but this did not substantially affect the yield of incidental findings detection [13–16]. Conversely, the heterogeneity of the studied populations might explain the variable prevalence. The percentage of current or former smokers is highly variable (respectively 6.7 to 42% and 18.3 to 39%) [8,14–19]. The percentage of diabetic patients is unknown in some cohort [13,15,19]; diabetics patients were excluded in the study of Haller et al. [17] and only 18% were included in the cohort of Onuma et al. [18]. In this last study, there is no report of a higher prevalence of incidental findings in diabetic patients. A systematic description of all incidental findings revealed in a diabetic population and their management was not performed in any study.

The higher prevalence of incidental findings in type 2 diabetes than in type 1 diabetes may be related to general insulin resistance and inflammation involved in the pathophysiology of type 2 diabetes, whereas type 1 diabetes is a pancreatic auto-immune disease. An increased cancer prevalence has been reported in both in type 2 diabetes and metabolic syndrome. The reported cancers were endometrial, breast, lung, pancreatic, hepatocellular and colorectal cancers [20,21]. Altough insulin has been suspected to be involved in a higher incidence of cancer by inducing an overexpression of growth factors, no association with the prevalence of incidental findings was found.

However, we also found in our cohort a higher occurrence of incidental findings in type 3 diabetes. These patients are more often in frailty conditions and may exhibits additional comorbidities contributing indirectly to the increased prevalence of incidental findings we found. The difference of incidental findings between type 3 and type 2 diabetes was not significant, probably by a lack of power with a small type 3 diabetes group. Nevertheless, only three lung cancers were identified in our cohort, all in type 2 diabetic patients. Lung cancer prevalence ranged from 0–1.2% in previous studies [13–19] leading to an overall prevalence of 0.18% which is not significantly lower than that found in our cohort (0.4%).

Unsurprisingly, active or weaned smoking is associated to incidental findings in our cohort. Most incidental findings identified are pulmonary and smoking is a risk factor for a variety of lung diseases apart lung cancers (chronic obstructive bronchitis, emphysema, fibrosis, bronchiectasis …) [22]. The association of microalbuminuria with both incidental findings and pulmonary nodules is unexpected and was not previously documented in literature. Nevertheless, microalbuminuria and diabetic nephropathy are associated with an activation of fibrosis, oxidative stress and chronic inflammation pathways, also involved in tumorigenesis [23,24]. Further investigations are needed to confirm and/or elucidate the potential link between micro-albuminuria and incidental findings in diabetic patients.

Regarding the care monitored in real life, almost half of the patients with incidental findings had an additional TDM, leading to an increased irradiation. However, this represents only 6.3% of the whole cohort. Recent studies reported a reduction of lung cancer mortality after regular screening by low-dose computed tomography (LDCT) in heavy smokers [25,26]. Since smoking is a common risk factor for both ischemic coronary disease and lung cancer risk, the question of performing a LDCT directly coupled with CAC in smokers and/or in patients with a higher cancer risk such as type 2 diabetic patients is open. Moreover, LDCT would increase the sensibility of detection of lung nodules and parenchymal lung abnormalities compared to MCDT performed for CAC assessment. Conversely, such a strategy could increase lung cancer overdiagnosis: a recent meta-analysis considered almost half of lung cancers detected by LDCT were over diagnosed [27].

In our study, the CAC and the subsequent TDM allowed an early diagnosis of lung cancer in three smoking patients who were still alive after two and three years of follow-up. Among the micronodules, none of them had significantly increased or became cancerous. However the duration of follow-up (on average 23 months) was too short in this study to determine whether this careful monitoring will be beneficial to these patients on a long term follow up.

Thus, the tradeoff between risks and benefits is important to consider in this population with a high CV risk. The benefit of an early identification of an incidental finding using MDCT provides a unique opportunity to detect an illness early- phase for which the outcome could be improved by early treatment. Conversely, several studies raise the issue of insignificant accidental discoveries, which in addition to anxiety, would lead to further exploration increasing irradiation without proof of a real benefit [12].

One limitation of our study is the monocentric design leading to a limited number of patients in the cohort, which makes statistical analysis less powerful. Additionally, the lack of rereading performed to check for oversights could be a second limitation. However most of the incidental findings were submitted to an independent specialized expert. Their consistency was established, and this situation corresponds to real life.

Furthermore, the duration of lung nodules follow-up was short because they require a demonstration of stability over at least 2 years and our study was only conducted during 23 months. The retrospective design precludes any bias of observation regarding the evaluation of the efficiency of care provided following CAC assessment. It represents a fair description of the care based on the electronic medical file with an excellent completeness since 98% of the items were collected in database. Few data is available in literature on the yield of the downstream care following CAC assessment.

Conclusions

To conclude, our findings raise the question of a need to combine CAC assessment with a LDCT, in order to explore a whole chest window with more sensibility, in type 2 and type 3 diabetic patients or in smokers, who are most likely to present incidental findings or cancers. Additional cohort studies comparing different options are needed to validate such a strategy.

Supporting information

Acknowledgments

The authors thank Hospices Civils de Lyon for their support, the imaging department (Dr Delphine Gamondès), the nurses of the endocrinology department of Hopital Louis Pradel and Caroline Moulin for her expert editorial assistance.