Approach to Acute Respiratory Illness in Children with Hematological Malignancy: A Prospective Study Evaluating Utility of CT Scan

Abstract Introduction  Various pulmonary complications can occur in children with hematological malignancies including both infection and malignant disease infiltration of pulmonary parenchyma. Objectives  To assess the role of CT scan in determining the etiology of acute pulmonary complications in children with hematological malignancies. Materials and Methods  All children < 17 years with newly diagnosed hematological malignancy with respiratory symptoms (Group A) along with children who developed fever with persistent respiratory symptoms as well as worsening chest radiographs during treatment (Group B) and underwent CECT thorax, from February 2019 to July 2020 were enrolled. The final diagnosis was made on the basis of clinical history, laboratory as well as radiological investigations and treatment response. Results  Thirty-seven children with mean age of 7.5 ± 3.5 years and male to female ratio of 1.3:1 who underwent CECT thorax were included in our study. For newly diagnosed cases, i.e., Group A ( n  = 8), the most common cause of respiratory symptoms as identified on CECT thorax was pulmonary tumoral infiltration ( n  = 5) followed by tuberculosis ( n  = 3). However, in Group B ( n  = 29) the cause of persistent respiratory symptoms was identified as infection ( n  = 17) followed by leukemic infiltration ( n  = 12). Thus, chest CT could accurately identify pulmonary tuberculosis, fungal pneumonia, bacterial infection, and pulmonary tumoral infiltrates. Conclusion  CT scan can be used as an adjunctive tool for prompt diagnosis and management of pulmonary complications in children with persistent respiratory symptoms as they are often non-specific.


Introduction
Acute respiratory illness (ARI) in immunocompromised children remains a significant diagnostic challenge for both clinicians and radiologists. Pulmonary infection complicates $75% of immunocompromised patients during their treatment. 1 Given the innumerable pathogens that can cause infectious complications in immunocompromised children, identifying a specific pathogen can be elusive.
In children developing complications secondary to chemotherapy-induced immune suppression, when combined with certain specific morphological findings and the type and duration of immune suppression, radiological investigations can aid with the decision-making process. Chest radiography is the initial imaging modality of choice for the diagnostic assessment of patients presenting with ARI; however, there is low sensitivity and low specificity regarding the type of specific pathogens. 2 Chest CT has been demonstrated to confer a higher degree of sensitivity and specificity for identifying the underlying cause of pulmonary involvement.
Contrast-enhanced CT chest should be done to evaluate the lungs, mediastinum, and pleural/chest wall abnormalities and identify mediastinal and hilar lymph nodes, as opposed to axial HRCT (high-resolution CT), which is done for pulmonary pathologies only. Volumetric multidetector CT acquisition in most of the modern day scanners can generate reconstructed HRCT images, which obviate the use of traditional axial HRCT acquisition in pediatric patients. Apart from pulmonary infections, non-infectious causes such as malignant pulmonary infiltration, radiationinduced lung injury, pulmonary thromboembolic phenomenon, should also be considered while evaluating acute respiratory illness in this group.
The aim of this study was to assess the role of contrastenhanced CT scan in determining the etiology of pulmonary complications in children with hematological malignancies presenting with ARI.

Materials and Methods
This was a prospective observational study performed at the Division of Pediatric Hematology Oncology, Department of Pediatrics, Institute of Medical Sciences, Banaras Hindu University from February 2019 to July 2020 (ethical clearance no. Dean/2018/EC/919). Thirty-seven children were included in this study who met our inclusion criteria.

Inclusion Criteria
All children < 17 years with hematological malignancy who were Newly diagnosed with respiratory symptoms (group A) On chemotherapy and developed fever with persistent respiratory symptoms, i.e., respiratory symptoms present even after 7 to 10 days from start of treatment along with/without worsening chest radiographs (group B)

Exclusion Criteria
Unwillingness to participate in the study Hemodynamically unstable child Children were investigated and empirical intravenous antibiotics, i.e., piperacillin-tazobactam and amikacin were started as per unit protocol. Investigations such as complete blood count, conventional blood culture, chest X-ray were sent for all children that had either of the two, i.e., fever, cough, or increased respiratory rate for age. Intravenous/oral fluconazole (antifungal) was added after 48 to 72 hours 3 if the child had clinical deterioration despite receiving empirical intravenous antibiotics. As pulmonary infection in immunocompromised children is most likely due to bacterial pathogens, 2 antibiotics was started and resolution of symptoms were expected to occur by 7 to 10 days (arbitrary) with antibiotics, followed by antifungals (as indicated) as per blood culture sensitivity pattern in the pediatric oncology ward. Patients responding to antibiotics in both the groups and having a normal follow-up chest X-ray at 7 to 10 days did not undergo chest CT. In both the groups, if after 7 to 10 days, there was no clinical improvement to empirical antibiotics, a repeat conventional blood culture, chest X-ray, and chest CTwas done. The indications of chest CT in our study were persistent respiratory symptoms, irrespective of abnormality on chest Xray at 7 to 10 days of antibiotics.
Chest CT findings were further corroborated by blood culture, gastric aspirate for Gene Xpert and galactomannan assay (based on availability) in infective cases. As gastric aspirate in children has been found to be equally effective and beneficial in Mycobacterium tuberculosis (MTB) detection with non-requirement for specific facilities as compared with bronchoalveolar lavage, it was used in this study for MTB detection. 3 CT-guided tissue biopsy from mediastinal mass was performed in most cases where malignant infiltration was suspected. Children whose parents did not consent for tissue biopsy, it was not done. Pulmonary infiltration was suspected if the clinical condition of the child was consistent with disease infiltration and child had sterile cultures. However, lung biopsy was not performed.
Data were collected with respect to age, gender, duration of symptoms, radiological findings on chest X-ray and chest CT, along with response to treatment. In infective cases, results of blood culture, gastric aspirate for Gene Xpert and galactomannan assay (based on availability) along with results of tissue biopsy in cases where malignant infiltration was suspected were also collected.

CECT Protocol
CT chest was performed using a 128 slice light speed VCT (GE Medical System). Non-ionic contrast agent Iohexol (Omnipaque) of concentration 300 mgI/mL was administered using hand injector (with an optional use of power injector and bolus chase) in a calculated dosage(1-1.5 mL/kg) depending upon the child's weight. In patients with previous history of contrast allergy or high-risk of contrast allergy (previous anaphylactic reactions to food or medication), after injecting the first 1 mL of contrast agent, child was observed for a few seconds for any allergic reactions and again after half an hour to see any delayed reaction. Scan delay of 25 to 30 seconds was kept following the administration of contrast for optimal enhancement of soft tissue. Images were then reconstructed using different reconstruction algorithm and evaluated in advantage workstation 4.4 and 4.7.

Image Evaluation and Interpretation
All CT scans were evaluated by a radiologist with 7 years of experience. Following definitions were used for characterizing lesions seen on CT scan. 4 i. Consolidation is defined as homogenous increase in lung parenchymal attenuation with obscuration of airways and vessels. ii. Ground glass opacities (GGO) are defined when bronchovascular margins are preserved behind hazy area of homogenously increased attenuation. iii. A nodule around the peripheral pulmonary arterial branches or 3 to 5 mm from the pleura, pulmonary vein or interlobular septa is defined as centrilobular nodule. Appearance of multiple areas of centrilobular nodules with a linear branching pattern was defined as "tree-in-bud" nodules. Nodules were termed as random if they lacked an architectural predominance. iv. Abnormal widening of the interlobular septa is defined as an interlobular septal thickening. v. Each parenchymal lesion localized to a particular lobe was evaluated in terms of presence and distribution of abnormality identified on CECT thorax. vi. Halo sign was defined as ground-glass opacity surrounding a nodule or, a mass or a rounded area of consolidation.

Approach to Diagnosis
Following signs on CT were used to suggest the specific etiology of pulmonary findings.
• Bacterial pneumonia: Patients with segmental or lobar consolidation with/without pleural effusion • Tuberculosis: Centrilobular nodules with tree in bud pattern, necrotizing mediastinal and hilar lymph nodes, empyema, cavitatory consolidations, military pattern, and cavitatory lesions. • Fungal infection: Patchy consolidations and halo sign, cavitatory nodules, consolidation with crescent of air. • Viral pneumonia: bilateral multifocal GGO with patchy consolidations with perihilar distribution along bronchovascular bundles indicative of bronchopneumonia. • Lymphoma/leukemic infiltrate: mediastinal mass with lung consolidation or randomly distributed nodule with bronchovascular bundle thickening or nodular pleural thickening All children were followed and a repeat CECT was performed after the completion of treatment in children with tuberculosis and malignant pulmonary infiltration. The final etiological diagnosis was made on the basis of clinical history, laboratory investigations and treatment response was considered as an outcome measure in this study.
Statistical Analysis: SPSS software (Version 22.0. IBM Corp. Armonk, NY, USA) for windows was used for data entry and analysis. All numerical variables are expressed as median with range. For comparison of categorical data chi square test was used. For categorical variables with cell values < 5, Fisher's exact test was used. A p-value < 0.05 was taken as significant.
Ethics: The procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional or regional) and with the Helsinki Declaration of 1964, as revised in 2013. Ethical clearance from Institute Ethical Committee of Institute of Medical Sciences, Banaras Hindu University, was obtained (Dean/2018/EC/919). Written informed consent was obtained from parents of all children included in our study.
Among children who underwent chest CT in both groups, 24.3% children (n ¼ 9) despite having a normal chest X-ray on day 7, underwent chest CT to elucidate the cause of respiratory distress. Chest CT was able to detect pulmonary changes, i.e., tubercular (n ¼ 2) in group A and disease infiltration (n ¼ 3), followed by pneumonia, i.e., fungal in (n ¼ 3) and bacterial (n ¼ 1) in group B.

Pulmonary Tuberculosis
CECT suggested pulmonary tuberculosis in seven children who demonstrated centrilobular nodules with tree in bud appearance (n = 6) most frequently (►Fig. 2A), followed by segmental/subsegmental consolidation (n = 4), and military tuberculosis with random nodules in one child. Other findings were pleural effusion (n = 3), GGOs (n = 1), and necrotic mediastinal lymph nodes (n = 3). Gene Xpert (gastric aspirate) was sent for all these children that was positive in three children (Group A [n = 1]; group B [n = 2]). Antitubercular treatment (ATT) was started in all these seven children, irrespective of Gene Xpert results. Five children improved following ATT, and two children died due to relapse of underlying malignancy (►Table 2).

Fungal Infection
Children with CT features of fungal infection (n ¼ 6) revealed segmental/sub segmental consolidation (n ¼ 4), GGO (n ¼ 3), halo sign (n ¼ 3), reverse halo sign (n ¼ 1), multiple cavitatory lesions (n ¼ 1) and pleural effusion (n ¼ 3) (►Fig. 2B, 2C). Fluconazole (empirical antifungal) was started in all children (n ¼ 37) if there was no improvement in fever after adding antibiotics for 48 hours. Of the six children with CT features of fungal disease, galactomannan was positive in two children and Candida was isolated from blood in two children. Among these six children, in five children, antifungals were escalated, i.e., amphotericin B (n ¼ 4) and voriconazole (n ¼ 1). All six children responded to antifungals.

Discussion
The results of our study prove that CECT of chest done at 7 to 10 days of onset of persistent respiratory symptoms can efficiently detect the pulmonary involvement by infective/tumoral infiltrates, even with the normal chest radiographs. Further, CT scan can provide clues toward underlying etiology, narrow down the differential consideration, and help initiate additional diagnostic measures. Chest radiography remains the initial imaging modality of choice for the evaluation of immunocompromised patients presenting with acute respiratory symptoms. However, radiographs can be normal in up to 10% of patients with lung pathology in immunocompromised patients. 2 Our experience also showed that even with normal lung radiographs, CT could detect the findings in nine patients (34%). The American College of Radiology (ACR) appropriateness criteria for imaging of chest in immunocompromised patients (both adult and pediatric) recommend CT with contrast as the next imaging modality with normal, equivocal, or non-specific chest radiographs or with those demonstrating multiple, diffuse, or confluent opacities. However, separate criteria do not exist for pediatric patients, especially regarding the time to obtain CT scan. Our study suggests that CT with contrast should be done if the child is not responding to empirical antibiotic therapy at 7 to 10 days of persistent respiratory symptoms.
Children with bacterial pneumonia (n ¼ 7) had lobar/sublobar consolidations, along with pleural effusion (n ¼ 5) on chest CT as has been also reported by Copley et al. Caution should be exercised when diagnosing bacterial infection as CT findings can overlap with various non-infective causes such as pulmonary hemorrhage and acute eosinophilic pneumonia in children. 5 Also, CT findings may overlap with atypical pneumonia including viral infections but the presence of ground glass opacification with consolidation, peribronchial thickening and lobular instead of   segmental distribution 4 favor viral pneumonia which was not seen among any child in this study. Also, the absence of cavitatory lesion, necrotic mediastinal lymph nodes and predominant tree-in-bud nodules pointed toward bacterial etiology. The presence of centrilobular nodule with tree in bud appearance, segmental/sub-segmental consolidation (especially in upper lobes), have been described to be associated with pulmonary tuberculosis in immunocompromised host, 5,6 as was also seen in this study. Similar CT findings can also be seen in Pneumocystis jiroveci, 7 invasive pulmonary aspergillosis, 7 mucormycosis, 8,9 and candidiasis. 8,9 However, nodules or areas of consolidation with a surrounding halo of GGO representing pulmonary hemorrhage are cardinal features when considering a possibility of fungal pneumonia in neutropenic children. 10 Sometimes, areas of cavitation representing infarction also develop in the neutrophil recovery phase, 11 that are more frequently observed in Candida than Aspergillus infection. In our study, one of six children had multiple cavitatory lesions that improved following fluconazole administration that favors the possible candida infection in this child.
The role of chest radiographs in detecting non-infectious complications such as tumor recurrence and pulmonary metastasis in children with history of malignancy, is limited. In this study, seven children who had disease infiltration (confirmed by CECT findings as well as clinical condition) showed inhomogeneous opacities on chest radiographs. The presence of pulmonary tumoral infiltration on CT scan is usually suggested by presence of a coexisting mediastinal mass (lymphoma) and lung consolidation or randomly distributed nodule with bronchovascular bundle thickening (leukemia), reflecting predilection of leukemic infiltrates for perilymphatic pulmonary interstitium or nodular pleural thickening 12,13 Thus, we would like to emphasize that chest radiography still remains the imaging modality of choice for the diagnostic assessment of immunocompromised children presenting with acute respiratory illness. 14 It cannot only detect infectious complications such as pleural effusion, empyema, and pneumothorax 14,15 but serial chest radiographs along with pattern and distribution of abnormality can also aid in formulating a differential diagnosis. 14,15 However, they lack sensitivity to detect subtle abnormalities in symptomatic immunocompromised children. Also, chest radiographs are non-specific with regard to specific pathogen. 1 On the contrary, chest CT is more sensitive than chest radiography for detecting subtle pulmonary parenchymal abnormalities due to its superior spatial resolution and cross-sectional display of findings. 14 In a study by Heussel et al, CT could aid in diagnosing pneumonia in 60% of febrile neutropenic patient with normal chest radiograph at least 5 days before abnormalities were visible on chest radiographs. 16 The results of our study emphasize that utilizing a chest radiograph as an initial modality and incorporation of CECT as a second-line modality is a cost-effective approach that can be utilized especially in low-income countries. Moreover, CT can obviate the need for specialized biochemical testing  Mediastinal widening Large mildly enhancing mediastinal mass with necrosis, nodular thickening of pleura with pleural effusion which can eventually decrease the overall cost incurred to the patients. The major limitation in this study was our small cohort size and lack of detailed microbiological/histopathological assessment through lung biopsy. In the majority of cases, our diagnosis was based on response to antibiotics and clinical follow-up. Also, we did not screen for viral infections due to a high cost of PCR based test to detect the same. The strength of this study was, however, its prospective nature. Therefore, we suggest that there is a need for multicentric prospective studies that can evaluate the role of CT scan in diagnosing etiology of persistent respiratory symptoms in pediatric immunocompromised patients in a resource limited setting.

Conclusion
In conclusion, chest CT can be a useful adjunctive tool when the etiology for ARI is not clear and resources are limited. Chest CT has the ability to identify the presence of pulmonary disease with high sensitivity, along with vital diagnostic information about the pattern of involvement as well as the most likely etiological pathogen.

Data, Materials and/or Code Availability
Data are available with the corresponding author and can be shared on reasonable request.

Ethics Approval
The study was approved by the institutional ethics committee of Institute of medical Sciences, Banaras Hindu University.

Funding
None.

Conflict of Interest
None declared.

Consent
Written informed consent was obtained from all the patients included in this study.