CT Features of Lymphatic Plastic Bronchitis in Adults: Correlation with Multimodality Lymphatic Imaging

Introduction

Plastic bronchitis (PB) is a rare condition defined by recurrent expectoration of bronchial casts, which are cohesive branching molds of the tracheobronchial tree. PB has been described in patients with respiratory disease, lymphatic abnormalities, viral infection, and congenital heart disease (1–4). Multiple classification systems have sought to characterize patients on the basis of their associated disease state (5), bronchial cast histopathologic features (6), or both (1). Advances in lymphatic imaging have identified abnormal pulmonary lymphatic flow from the thoracic duct into the bronchial tree as an important cause of PB (7,8). The term lymphatic plastic bronchitis was coined, with some authors now advocating for the classification of patients as having lymphatic or nonlymphatic PB (8,9). The correct diagnosis of lymphatic PB is critical to allow for curative percutaneous lymphatic intervention (8,10). There is growing evidence that lymphatic PB is the predominant cause of PB, which may occur in patients with congenital heart disease (4) and in adult patients (11,12). A recent review of the literature identified 31 adult patients with PB, of which 39% (12 of 31) had abnormal lymphatic flow at lymphangiography (11). However, the true prevalence of PB and its subtypes is unknown due to its rarity and likely frequent misdiagnosis (13).

Patients with PB can present with a variety of symptoms, including dyspnea and recurrent respiratory tract infections prior to or in combination with the defining symptom of cast expectoration. Overlapping symptoms can confound true PB with respiratory diseases that cause mucus plugging (eg, allergic bronchopulmonary aspergillosis, asthma). Patients will typically undergo clinical and laboratory testing for asthma, allergic bronchopulmonary aspergillosis, autoimmune conditions, and immunodeficiency. Bronchoscopy aids in identifying and removing bronchial casts. There is no standardized workup given the rarity of PB and the heterogeneous group of potential causes. Chest CT is important to identify alternative causes of respiratory disease, although it has limited specificity in helping to identify the cause of plastic bronchitis (3). Little has been described of the CT findings of patients with PB. The largest case series of 34 pediatric patients reported “variable radiologic findings without diagnostic significance to PB” (14). Atelectasis, either unilateral or basilar segmental, was the most common finding reported in their sample of patients with single-ventricle physiology or associated respiratory disorders. Reports of adult patients with PB are limited to case reports or small case series (8,15,16).

Identifying patients with lymphatic PB has become a key decision point in their treatment. To date, nonspecific chest radiology findings of PB have been reported. Our goal was to identify CT patterns to distinguish between lymphatic and nonlymphatic causes of PB.

Materials and Methods

Patient Selection

Institutional review board approval was obtained, and written informed consent was waived given the retrospective nature of this study. Review of a prospectively maintained Health Insurance Portability and Accountability Act–compliant institutional database identified 56 consecutive patients with PB who presented for lymphatic imaging at a single hospital between January 2014 and August 2020. This hospital has a specialized center for lymphatic intervention, with a national and international referral base. Patients were excluded if they did not undergo successful lymphatic imaging (n = 1), had previously undergone thoracic duct embolization and/or ligation (n = 5), or chest CT images were unavailable (n = 6) (Fig 1). A retrospective review of imaging findings and clinical data was performed in 44 patients. Patients with retrograde flow of contrast media into the lung parenchyma at lymphatic imaging (either MR or fluoroscopic lymphangiography) were defined as lymphatic PB.

Figure 1:

Download as PowerPoint

Chest CT Acquisition and Analysis

Chest CT equipment and protocols were inconsistent due to the broad referral pattern. The protocols included multiplanar unenhanced (n = 28) or contrast-enhanced images (n = 16) of section thickness ranging from 1.0 to 5.0 mm. Lung kernel reconstructions were unavailable for six patients. Thin section (1.0-mm section thickness) imaging was available in 19 patients, 5.0-mm section thickness was available in seven, and the remainder (n = 18) had 2.0–3.0-mm section thickness. All image review was performed with Sectra Workstation IDS7 (Version 20.2.14.3442; Sectra AB). The CT images were anonymized, randomized, and reviewed independently by two fellowship-trained thoracic radiologists (M.H. and L.R., 1 year of experience each). The readers were blinded to all clinical information, including subsequent lymphatic imaging findings.

The CT findings were interpreted on the basis of the recommendations of the nomenclature committee of the Fleischner Society (17). Disagreements were resolved by blinded independent review by a senior thoracic radiologist with 5 years of experience (M.G.A.).

Radiologists evaluated the lobar distribution of ground-glass opacity (GGO), recording the most severely affected lobe and noting a semiquantitative assessment of the extent of GGO as 0%–33%, 34%–66%, or 67%–100%. GGO was classified as bronchocentric (occurring axially along the bronchovascular interstitium), peripheral (a predominance for the outer one-third of the lung or along the interlobar fissures), or random (no distinct bronchocentric or peripheral predominance). Consolidation, interlobular septal thickening, and bronchial wall thickening were defined as diffuse, patchy and related to GGO, or patchy and unrelated to GGO. Mucoid impaction and bronchiectasis were assessed for upper or lower predominance and anatomic distribution: central, segmental bronchi, or random. The presence and location of airway casts (eg, cohesive filling defects of the large airways) were recorded. Mediastinal and hilar lymph nodes were rated as normal or enlarged (short axis > 1 cm).

Lymphatic Imaging

An ideal lymphatic workup began with dynamic contrast-enhanced MR lymphangiography (DCMRL) followed by thoracic duct lymphangiography (TDL). In brief, both techniques involved the injection of contrast medium into the inguinal lymph nodes: oil-based contrast medium (Lipiodol; Guerbet) for TDL and gadolinium-based contrast medium for DCMRL. The progression of contrast medium through the central conducting lymphatics was assessed using fluoroscopy and a 1.5-T MRI machine, respectively, using previously described methods (18,19). DCMRL images were available in 33 of 44 patients. DCMRL images were unavailable due to body habitus preventing DCMRL (n = 5), critical illness preventing DCMRL (n = 1), pacemaker preventing DCMRL (n = 1), and unavailable images from an outside hospital (n = 4). TDL was performed in 41 of 44 patients. Three patients did not proceed to TDL after the TD was identified as normal at DCMRL. This was based on our initial experience of six patients with normal TD at DCMRL that was confirmed as normal at TDL. TDL was performed simultaneously with bronchoscopy and injection of methylene blue dye in the TD in 13 patients. Bronchoscopy confirmed abnormal submucosal bronchial perfusion in patients with questionable DCMRL findings or facilitated bronchoscopic removal of casts in patients who had required previous cast removal. Unblinded review of the DCMRL and TDL images was performed by the interventional radiologist (M.I.) who performed the procedures and has 20 years of experience in lymphatic imaging. Presence and location of pulmonary lymphatic perfusion (PLP) in the mediastinum and/or lung parenchyma were recorded for both modalities as follows: (a) hilar PLP; (b) neck PLP, if it originated at or above the clavicles; and (c) diaphragmatic PLP, if it originated at or immediately above the diaphragm. TDL images assessed the TD anatomy and patency. CT findings were compared with lymphangiographic findings at DCMRL and/or TDL, and with bronchoscopic dye evaluation, when performed.

Statistical Analysis

Statistical analyses were performed in Microsoft Excel 2020 (version 16.36) and the R statistical environment (version 3.6.3; R Foundation for Statistical Computing) by author Q.C. Demographic and procedural variables were summarized by using standard descriptive statistics. Pearson correlation coefficient (r) was used to assess the relationship between frequency of cast production and severity of GGO. Interobserver variation in CT findings was quantified by using the unweighted κ coefficient of agreement for presence variables (20). The interobserver agreement was classified as follows: poor, less than 0; slight, 0–0.20; fair, 0.21–0.40; moderate, 0.41–0.60; substantial, 0.61–0.80; and almost perfect, 0.81–1.00 (21). Categorical end points were compared with Fisher exact test. A P value less than .05 indicated statistical significance. Univariate logistic regression was used to evaluate the relationship between GGO and lymphatic PB. Receiver operating characteristic curve analysis was conducted to assess the predictiveness of GGO; area under the receiver operating characteristic curve (AUC) values were reported. Sensitivity and specificity of GGO were calculated by comparing the CT findings with lymphatic imaging outcome.

Results

Patient Demographics

Patient demographics and indication for lymphatic imaging are summarized in Table 1. Lymphatic PB was determined in 35 patients and nonlymphatic PB in nine patients. Of the patients in the nonlymphatic group, an alternative cause of PB was identified in seven of nine patients, namely asthma and/or hypereosinophilia (n = 3), sarcoidosis (n = 2), and postinfectious bronchiectasis (n = 2). Two patients did not have an alternative diagnosis.

CT Findings

CT findings and interobserver agreement are presented in Table 2. Dichotomous presence variables had substantial agreement (0.61–0.80), while the remaining ordinal variables had agreement ranging from fair (0.21–0.40) to substantial (0.61–0.80).

For the entire group of patients with PB, multifocal GGO was the predominant imaging finding (72.7% [32 of 44]). There was substantial agreement of the pattern of GGO (κ, 0.61; 95% CI: 0.39, 0.84) and the lobes involved (κ range of 0.65–0.82). Distribution was bronchocentric in 27.3% (12 of 44) of patients, peripheral in 11.4% (five of 44), and random in 40.9% (18 of 44), with fair interobserver agreement (κ, 0.35; 95% CI: 0.12, 0.57). Consolidation was present in 38.6% (17 of 44), of which 11 were associated with GGO and six were not. Smooth interlobular septal thickening associated with GGO was observed in 31.8% (14 of 44) of patients and created a “crazy paving” pattern of varying severity. One patient in the nonlymphatic PB group had nodular interlobular septal thickening without GGO that was attributed to the patient’s sarcoidosis. Features of airway obstruction (either atelectasis and/or mucoid impaction) were present in 59.1% (26 of 44) of patients. Mucoid impaction was located in the segmental airways in 18.2% (eight of 44), randomly in 9.1% (four of 44), and centrally in 6.8% (three of 44). Bronchiectasis was present in 13.6% (six of 44) of patients. Pleural effusions were present in 9.1% (four of 44) of patients. One patient with lymphatic PB had tortuous mediastinal vessels that proved to be lymphatic vessels at lymphangiography.

Patients with lymphatic PB were more likely to have GGO (Fisher exact test, 91.4% [32 of 35] vs 33.3% [three of nine]; P < .001) (Fig 2). Univariate logistic regression confirmed this result by showing that GGO was a significant predictor in the model (odds ratio, 21 [95% CI: 3.83, 158.73]; and risk ratio, 2.74 [95% CI: 1.08, 6.95]). The model AUC of GGO was 0.79 (0.62–0.96), further reassuring that GGO is predictive of lymphatic PB in this patient cohort. GGO was more commonly observed in the lower lobes, 88.6% (31 of 35), compared with the upper lobes, 68.6% (24 of 35). Compared with patients with nonlymphatic PB, patients with lymphatic PB had more involvement in the right middle lobe (65.7% [23 of 35] vs 11.1% [one of nine]; P = .015) and the right lower lobe (80% [28 of 35] vs 22% [two of nine]; P = .004). There was correlation between GGO in the right middle and right lower lobes (Cramer V = 0.60). The frequency of cast production did not correlate with the percentage of GGO in the worst affected lobe or the number of lobes involved (r = 0.056 and −0.12, respectively). Regarding the 8.5% (three of 35) of patients who did not have GGO at their initial CT, the first was asymptomatic at the time of their CT, and the second patient underwent their CT at the onset of small cast production. As their symptoms progressed, subsequent CT images showed GGO in their right middle and lower lobes. The third patient had smooth right lower lobe interlobular septal thickening without GGO that resolved after embolization. Features of airway obstruction (atelectasis and/or mucoid impaction) were present in 51.4% (18 of 35).

Figure 2:

Download as PowerPoint

Airway Casts

Cohesive, branching airway casts were identified in 11.4% (five of 44) of patients, each who had lymphatic PB. In each patient, these casts were observed in the airway supplying a lobe or segment affected by GGO, namely the right superior lobar bronchus (n = 2), right inferior lobar bronchus (n = 1), lingular bronchus (n = 1), and bronchus intermedius, right superior lobar, and right inferior lobar bronchi (n = 1).

Lymphangiographic Correlate

Table 3 compares the location of GGO with PLP (Figs 3, 4). The mean time from initial CT to lymphatic imaging for all patients was 233.4 days (range, 1–1350 days) and 130 days (range, 40–199 days) for the group of patients with nonlymphatic PB. The sites of GGO at CT and PLP at lymphatic imaging were discordant in 22.9% (eight of 35) of patients: four patients with bilateral GGO and left hilar PLP, three patients with right-sided GGO and bilateral hila with left neck PLP, and one patient with right-sided GGO and left hilar and diaphragmatic PLP. Bronchoscopy and methylene blue dye injection were performed during TDL in 29.5% (13 of 44) patients. The location of methylene blue dye leak was concordant with lymphatic imaging and CT findings in 61.5% (eight of 13) (Fig 4). Three of the discordant patients had additional areas of GGO contralateral to the methylene blue dye leak, one patient had left mainstem bronchus leak without left-sided GGO, and one patient had no methylene blue dye leak with perihilar perfusion at DCMRL and bilateral GGO at CT.

Figure 3:

Download as PowerPoint

Figure 4:

Download as PowerPoint

Follow-up Data

Symptoms improved or resolved after lymphatic intervention in all patients with lymphatic PB.

Test Characteristics

For the group of patients presenting with PB, the sensitivity of GGO for lymphatic PB was 91% (32 of 35; 95% CI: 76, 98) and specificity was 66% (six of nine; 95% CI: 30, 93). Predictive values were not calculated because of the low prevalence of PB.

Discussion

PB is a rare condition defined by the expectoration of bronchial casts. The true prevalence of PB is unknown due to its rarity and likely frequent misdiagnosis. A wide variety of systemic and pulmonary illnesses have been associated with PB, including asthma, allergic bronchopulmonary aspergillosis, bronchiectasis, and cystic fibrosis (2,8,14). Lymphatic abnormalities have a strong association with PB in patients with single-ventricle heart disease and lymphatic flow disorders (2,10). This important subset of patients has retrograde flow of lymphatic material that leaks and congeals within the bronchial tree. It is thought that lymphatic anomalies, likely congenital, are triggered by bronchial injury or increased lymphatic flow (22,23). Identifying patients with lymphatic PB allows for potentially curative lymphatic intervention. This study of chest CT findings in adult patients referred for lymphatic workup of PB found that GGO was more likely in patients with lymphatic PB compared with other causes of PB (91.4% [32 of 35] vs 33.3% [three of nine]; P ≤ .001). These authors advocate that the presence of GGO on chest CT images in a patient with PB, in the absence of confounding acute diagnoses, should prompt a referral for lymphatic imaging. Importantly, chest CT should be performed when the patient is symptomatic. Two of three patients with lymphatic PB without GGO at their initial chest CT were asymptomatic at the time of imaging and subsequently developed GGO as their symptoms progressed.

These results advance the currently limited role of chest CT findings in the workup of PB. The existing literature concerning radiographic findings of PB is limited to case reports and case series that report nonspecific findings of airway obstruction, including mucoid impaction, atelectasis, and nonspecific opacities (6,14,24,25). GGO and interlobular septal thickening have been reported in isolated case reports of pediatric and adult patients with underlying pulmonary lymphangiectasia (26,27). By contrast, 91.4% (32 of 35) of the patients with lymphatic PB had GGO at CT, while 51.4% (18 of 35) had features of airway obstruction (mucoid impaction and/or atelectasis). GGO is a nonspecific finding that indicates partial filling of alveoli or interstitial thickening from a wide variety of conditions (eg, pulmonary edema, infection, alveolar hemorrhage, and others).

We hypothesize that GGO in our study population represents partial filling of alveoli with overflowing lymphatic fluid, dilated lymphatic channels, or a combination of both. The increased prevalence of GGO in the patients with lymphatic PB supports its association with abnormal pulmonary lymphatic perfusion. In addition, GGO corresponded to the location of lymphatic imaging abnormality or abnormal submucosal perfusion of methylene blue dye in the majority of patients. Discordance of GGO and lymphangiographic or methylene blue dye perfusion may be caused by fluctuance in symptoms over the occasionally long duration between CT and lymphatic imaging, by additional lymphatic pathways that were below the sensitivity of MRI and/or fluoroscopy, or by confounding pathologic features present at the time of CT.

Additional CT findings that were associated with GGO in patients with lymphatic PB were smooth interlobular septal thickening, bronchial wall thickening, and enlarged mediastinal lymph nodes. Interlobular septal thickening was associated with GGO in a third of cases, creating varying degrees of crazy paving. Of additional importance was the rarity of bronchiectasis and pleural effusions in patients with lymphatic PB. In future analyses, their presence may suggest an alternative cause of the PB or the presence of confounding pathophysiology at CT. The findings of GGO, crazy paving, and others are nonspecific and have a wide differential (eg, atypical pneumonia, pulmonary edema, diffuse alveolar hemorrhage, and aspiration bronchopneumonia). These more common conditions should be suggested initially. However, this study is of relevance when considering the specific presentation of chronic cast expectoration or removal via bronchoscopy. The presence of GGO should encourage the provider that lymphatic workup is appropriate.

The study had several limitations related to its retrospective nature and small size. The study population consisted of patients referred or self-referred to a tertiary lymphatic imaging center. This biases toward patients with positive lymphatic imaging and more severe symptoms, resulting in our small comparison group. These factors contributed to fair and/or moderate interobserver agreement for many imaging variables, high odds ratios, and wide CIs for our specificity and regression analyses. Additional factors that may have played a role included the lack of experience of our CT readers and heterogeneity in CT protocols. Many of these limitations are a factor of the rarity of PB and wide referral base, noting that our patient cohort was comparatively large compared with published data (14,28,29). Last, our study was focused on the sensitivity of GGO and role in selecting this specific group of patients for lymphatic imaging. The long duration between lymphatic and CT imaging may explain some of the discordant findings.

In all, we assessed CT findings in a comparatively large population of patients with PB. Our results identified GGO, often in combination with interlobular septal thickening, as a sensitive chest CT finding in patients with lymphatic PB. GGO is a nonspecific finding in chest radiology. However, in the clinical scenario of the patient expectorating casts, referral to a lymphatic imaging center for evaluation and potential treatment should be considered.

Disclosures of conflicts of interest: C.O. No relevant relationships. M.I. No relevant relationships. L.R. No relevant relationships. S.K. National Institutes of Health grant, unrelated to this article; National Institute of Biomedical Imaging and Bioengineering grant, unrelated to this article; imaging consultant to Trizell, unrelated to this article; consulting fees from Trizell as imaging consultant on a mesothelioma clinical trial; institutional support for travel from Penn; member of Radiology: Cardiothoracic Imaging editorial board. Q.C. No relevant relationships. M.H. No relevant relationships. M.G.A. No relevant relationships.

Author Contributions

Author contributions: Guarantors of integrity of entire study, C.O., S.K., Q.C.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, C.O., M.I., S.K., M.G.A.; clinical studies, C.O., M.I., L.R., S.K., M.H., M.G.A.; statistical analysis, S.K., Q.C.; and manuscript editing, all authors