Computed Tomography Characterization of Emphysema in Combined Pulmonary Fibrosis and Emphysema (CPFE) and its Correlation with Lung Cancer with Comparison to Chronic Obstructive Pulmanory Disease (COPD)

Purpose: To investigate imaging features of emphysema on high-resolution computed tomography (HRCT) in combined pulmonary brosis and emphysema (CPFE) and its correlation with lung cancer, with comparison to chronic obstructive pulmonary disease (COPD). Methods: 160 consecutive patients (80 CPFE and 80 COPD ) were analyzed retrospectively. The HRCT imaging features of the two groups were compared, and the correlation between factors underlying CPFE and the occurrence of lung cancer was analyzed. Results: We found paraseptal emphysema to be the more common form of emphysema in CPFE (38.75%), with an incidence statistically higher than that of COPD (χ²=14.462, p=0.001). The extent of emphysema in each lung lobe in the CPFE group was statistically less than that in the COPD group (p<0.01). Homogeneous emphysema was common in both the CPFE (73.8%) and COPD groups (52.5%), and it showed a higher incidence in CPEF group (χ²=7.760, p=0.004). Homogeneous emphysema also showed signicantly high correlation to lung cancer (χ²=9.298, p=0.004). The incidence of lung cancer in the CPFE group (20%) was higher than that in the COPD group (8.8%) (χ²=4.113, p=0.035). There was a linear relationship between the incidence of lung cancer and the degree of pulmonary interstitial brosis and years of smoking in CPFE patients (χ²=6.679, p=0.01* ,b=0.148). Conclusion: Different emphysema features showed between CPFE and COPD, and CPFE showed higher incidence of lung cancer. Homogenous emphysema, severity of damage caused by interstitial brosis and the number of smoking years have high relationship with incidence of lung cancer in CPFE.


Introduction
Combined pulmonary brosis and emphysema (CPFE) is a clinical entity that has been increasingly diagnosed over the last two decades [1,2] and is broadly characterized by upper lobe emphysema and lower lobe brosis [3,4]. The upper lobe emphysema contributes to a static volume and late phase in the forced volume test. In contrast, the lower lobe brotic portion contributes to early phase ow in the ow volume curve, and the KL-6 value (a blood marker of pulmonary brosis) is usually highly elevated [5].
The co-existence of pulmonary brosis and emphysema is considered to be an independent syndrome having a speci c disease progression, and CPFE is frequently complicated by pulmonary hypertension, acute lung injury, and lung cancer, especially non-small cell lung cancer (NSCLC) [6][7][8]. Previous studies have shown the severity of brosis have correlation with prognosis in CPFE. Prognosis is generally poor, and studies have shown that forced vital capacity (FVC) and brosis scores (FS) can be used to predict survival in biopsy-con rmed CPFE [9,10]. However, few studies have explored the correlation between emphysema features and lung cancer in CPFE.
The two most prevalent forms of emphysema in CPFE are paraseptal emphysema, and bullae, whose prevalence rates among CPFE patients have been reported to be 93 %, and 54 %, respectively [11]. Clinically, emphysema was also commonly seen in chronic obstructive pulmanory disease (COPD), and centrilobular emphysema showed more common in COPD [9,12]. However, few studies [13] have explored the detailed characterisitics of emphysema between the two and compared the incidence of lung cancer.
Therefore, for better differentiation in CPFE and COPD and deeper understanding in prognosis in CPFE, in this retrospective cohort study, we investigated the emphysema features including types, volume and uniformity in CPFE, with COPD as comparison, and we also investigated the possible factor correlating to lung cancer including emphysema types and uniformity with multi-factor statistical analyses.

Patient population
This retrospective study was approved by the local ethics committee board (Peking University First Hospital Ethics commitment, and the number of ethics was 2017[1382]) with a waiver of written informed consent. The methods were carried out in accordance with the Declaration of Helsinki.
A total of 160 patients diagnosed with either CPFE or COPD by a multiple team discipline (MDT) of radiologists and pulmonologists from Jan 2013 to May 2019 were included in this study. The CPFE cohort (80 out of 160) was the observation group, and the COPD cohort (80 out of 160) was the control group. Exclusion criteria included any other respiratory diseases, missing information, poor image quality, or other factors resulting in insu cient diagnostic con dence. The smoking history was recored for every participant.

Diagnostic criteria
We used CT-based diagnostic criteria to diagnose emphysema and pulmonary brosis.
Emphysema is characterized as having cystic lung lesions with low-density shadows and a clear boundary. There is no wall structure and only a thin visible wall (< 1mm), and the diameter of the pulmonary bullae is greater than 1 cm. Pulmonary brosis is characterized as having grid and honeycombing shadows predominating in the peripheral pulmonary and lower pulmonary areas; stretch bronchiectasis with a small amount of local ground glass opacity and/or consolidation is also seen. All the diagnoses were made by two senior radiologists who were expertise in pulmonary eld in consensus.
Both groups of patients underwent HRCT examinations, using a full-body CT scanner (manufacturer GE). The following scanning parameters were used: voltage 120 KV, current automatic adjustment, pitch 1.375, slice thickness 1.25 mm, slice spacing 1 mm. Patients were in a supine position throughout the scan. Breath-holding training was routinely given to patients before scanning. The scan coverage was from the thoracic inlet to the top of the diaphragm.
All the patients were undergone pulmonary functional tests, and the nal diagnoses of CPFE and COPD were con rmed by the MDT of sophisticated radiologists and pulmonologists Emphysema volume CT post-processing technology was used in the determination of emphysema volume. Thoracic VCAR is a non-invasive CT image analysis software package, which may be used in conjunction with CT lung images to aid in the assessment of thoracic disease (emphysema, COPD) diagnosis and management. In this study, thoracic VCAR software was utilized to obtain the emphysema volume of the left and right lungs and each lung lobe. When the difference in the volume percentage of any one of the lung lobes was greater than 10%, it was de ned as heterogeneous emphysema (Figure 1a). If this was not the case, it was de ned as homogeneous emphysema (Figure 1b).
The degree of interstitial brosis was characterized in the three groups based on the imaging criteria, including ground-glass opacity, thickened lea et interval, grid shadow, honeycomb lung, and stretch bronchiectasis ( Figure 2).

Statistical analyses
Data were analyzed using SPSS program version 18.0. The patient demographic data were analyzed rst to identify any statistical differences between the CPFE and COPD groups. Emphysema volume and emphysema types with and without the presence of lung cancer were compared between these groups. In addition, the relationship between multiple factors (gender, age, smoking status, emphysema volume uniformity, emphysema type, pulmonary interstitial brosis, and years of smoking) and the presence of lung cancer was analyzed.
All measurements are presented as average ± standard deviation (X±S), based on the descriptive statistics. All data evaluate normal distributions. Counting data is expressed as percentages (%). Chi-square tests were performed, and a p<0.05 indicates that the differences were statistically signi cant. A Chi-square test of association was used to evaluate the linear correlation of pulmonary interstitial brosis and the presence of lung cancer or smoking status in the CPFE cohort.

Demographic data
For the CPFE observation cohort, there were 72 males and 8 females, and the average age were (70.5±9.6) years (ranging from 48 to 88 years). For the COPD control group, there were 74 males and 6 females, and the average age were (72.6 ±11.2) years (ranging from 45 to 101 years). There were no statistical signi cance in the demographic differences between the two groups (Supplementary 1).

Comparison of CPFE and COPD cohorts: Emphysema types characterized by HRCT
The most common type of emphysema in the CPFE cohort was the paraseptal type, followed by the mixed type, and the incidences for both types were statistically higher than for those in the COPD group. The COPD was dominated by the centrilobular type at a statistically higher level than for the CPFE group (p<0.05) (Table 1, Figure 3).

Comparison of CPFE and COPD cohorts: Emphysema volume and uniformity
The emphysema volumes of both the left and right lung were signi cantly lower in CPFE group (p<0.01), and also showed signi cantly lower in each lung lobe in CPFE group (p=0.001), compared with the COPD group.
In the CPFE group, 59 cases (incidence of 73.8%) were homogeneous emphysema, while 21 cases were heterogeneous. In the COPD group, 42 cases (incidence of 52.5%) were homogeneous emphysema, while 38 cases were heterogeneous. Thus, homogeneous emphysema had a higher incidence in CPEF group. The Chi-square test also con rmed that the incidence of homogeneous emphysema in the CPFE group was signi cantly higher than that of COPD group (χ²=7.760, p=0.004 < 0.01) ( Table 2). Correlation between multi-factors and lung cancer in the CPFE group: Lung cancer incidence and characteristics in CPFE patients The analysis of correlation between multi-factors and lung cancer incidence revealed a lack of gender difference was observed (p>0.05). Age, smoker status and emphysema types did not have signi cant impact on lung cancer incidence (p>0.05), but the uniformity of emphysema correlated with lung cancer incidence. Speci cally, homogeneous emphysema patients were more likely to develop lung cancer (42.6%) than were the patients with heterogeneous emphysema (11.9%). (Table 3) Correlation between multi-factors and lung cancer in the CPFE group: The correlation of lung cancer to pulmonary interstitial brosis and smoking history There was a linear relationship between the incidence of lung cancer and the degree of pulmonary interstitial brosis and years of smoking in CPFE patients. The incidence of lung cancer increased with the severity of interstitial brosis and the duration of smoking. For every severity degree increase of interstitial brosis or every 10 years of smoking history, the cancer incidence rose by 14.8% or 14.1%, respectively. (Table 4, Supplementary 2).

Discussion
In our study, we investigated the detailed imaging features of emphysema in CPFE, with comparsion to COPD, and our results showed paraseptal emphysema was the most common form in CPFE, with an incidence statistically higher than that of COPD (p=0.001). Homogeneous emphysema was common in both the CPFE (73.8%) and COPD groups (52.5%), and it showed a higher incidence in CPEF group (p=0.004). The extent of emphysema in each lung lobe in the CPFE group was statistically less than that in the COPD group (p<0.01).
Due to the histopathological heterogeneity of CPFE, researchers often rely on HRCT for the diagnosis of CPFE, rather than using a pathological diagnosis [14][15][16][17]. HRCT can be used to visualize early changes in emphysema, as well as the in-depth details of interstitial pneumonia. HRCT thus plays a fundamental role in diagnosis of CPFE syndrome. Choi et al. [17] suggested the following diagnostic criteria based on HRCT: an area of emphysema with a reduced translucency clearly demarcated from adjacent normal lung tissue with a very thin wall (<1 mm) or no wall, and/or multiple pulmonary bullae (>1 cm) in the upper lobe. Centrilobular emphysema is a long-term, progressive lung disease and is considered to be a form of COPD, and studies [9,12] have shown that centrilobular emphysema and/or bullous emphysema occurring predominantly in the upper lung is seen as a focal low-density area clearly demarcated from normal lung tissue, with visibly thin walls (<1 mm) or no wall.
Paraseptal emphysema (seen as low-attenuation areas in the subpleural zone) has been described in a majority of CPFE reports [1], and Araki [11] suggest that paraseptal emphysema is a typical feature of CPFE. However, determination of the emphysema type in CPFE is often very di cult because of alteration of its features due to coexisting brosis [18]. In spite of this di culty, the presence of thick-walled large cysts-2 cm or more in diameter and delimitated by a wall 1 mm or more in thickness-in an area of the lung where reticulation is present, is considered to be one of the characteristic features of CPFE [19]. Based on the analysis of emphysema types in CPFE patients, this current study revealed that paraseptal type emphysema is the most common, a nding consistent with other studies [11,13]. And we also used COPD as control to further pove the emphysema differences between the two enteties.
In our study, by using the thoracic VCAR post-processing software, we found that homogeneous emphysema was the main type in CPFE patients. This research has not been conducted before between CPFE and COPD. In addition, the emphysema volume was lower than that of COPD patients. This may be due to the different pathological mechanisms of emphysema formation by the two diseases. Although both can lead to poor alveolar elasticity, valve action caused by chronic bronchiolitis might also contribute to the pathogenesis of COPD. A larger sample study is required to con rm this hypothesis.
This study also explored the incidence of lung cancer in CPFE patients. Our results show that CPFE patients are more likely to develop lung cancer, compared with COPD patients, with squamous cell carcinoma being the most commonly seen. These results are consistent with previous studies published by Kwak et al. [20]. Upon analyzing the relationship between multiple factors of CPFE and lung cancer, we found that lung cancer incidence is higher in homogeneous emphysema patients and increases both with the severity of pulmonary interstitial brosis and the number of years of smoking. We speculated that homogeneous emphysema, severe brosis, and length of smoking history are the principal risk factors for lung cancer in CPFE patients. The diffuse pulmonary interstitial brosis seen in CPFE often presents as reticular blurry shadows, honeycomb patterns, distortion of alveolar structures in the subpleural and lower lobe, and/or traction bronchial or bronchiolar dilation [13,19]. Focal ground glass opacity and/or alveolar consolidation may also occur but is not the primary manifestation. Occasionally, ground-glass opacity can be the only evidence of interstitial lung disease, and in this situation, lung biopsy becomes necessary [21].
In a few cases, pulmonary nodules and consolidation have also been found in CPFE, ndings which are correlated with a higher incidence of lung cancer [20].
Previous studies have shown that smoking is closely related to the occurrence of CPFE and lung cancer.
However, further research is needed to describe the correlation of homogeneous emphysema, degree of brosis, and lung cancer incidence. This information would contribute to both earlier detection of lung cancer and earlier treatment of patients, making increased survival of CPFE patients possible.
The limitations of this study include small sample size, and lack of in-depth characterization of interstitial brosis. In future study, we should conduct further investigation.

Conclusions
Pulmonary interstitial brosis with emphysema (CPFE) features unique imaging characteristics. Different emphysema features showed between CPFE and COPD, and CPFE showed higher incidence of lung cancer.
Homogenous emphysema, severity of damage caused by interstitial brosis and the number of smoking years have high relationship with incidence of lung cancer in CPFE. Radiologists treating patients with CPFE should be alert to the high incidence of lung cancer in this population. Clinical manifestations and imaging should also be included when managing CPFE patients. ). The need for written informed consent was waived by the institutional review board due to the retrospective design of the study.
The methods were carried out in accordance with the Declaration of Helsinki.

Consent for publication
All the authors consent for the publication.
Availability of data and material The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Competing interests
There is no competing interests among authors. Yan Xiong: Visualization, Investigation.