Cytomegalovirus (CMV) is one of the most important causes of opportunistic lung infection in the pediatric population [1
]. Risk factors include human immunodeficiency virus (HIV) infection and immunocompromised status after bone marrow/stem cell transplantation (BMT/SCT) and solid organ transplantation [2
]. Despite being a life threatening condition, if diagnosis is made in a timely manner, CMV pneumonia can potentially be cured with appropriate antiviral therapy (i.e.
, ganciclovir) [3
]. Unfortunately, the clinical diagnosis of CMV lung infection is challenging in children and often requires a high-index of suspicion, especially in non-HIV infected cases and patients not severely immunocompromised [4
Most of the literature about CMV lung infection in children focuses on the epidemiology of this condition in the pediatric immunocompromised population [1
]. There are few reports that address the specific clinical and pulmonary imaging findings that may suggest this difficult diagnosis. Smith et al.
in 1977 described the clinical respiratory findings and radiological appearance of CMV pulmonary infection in children using plain X-ray films [5
]. This seminal work enhanced the awareness of the atypical course of CMV infection in the lungs, often described as “CMV pneumonitis” [5
], which includes non-specific pulmonary findings such as interstitial patterns with increased bronchopulmonary markings and bronchiolar disease (diffuse air trapping) [5
]. Interestingly, there has been a dramatic change in both the epidemiology of CMV infection and diagnostic techniques available for its diagnosis in children over the last 30 years [6
]. Now-a-days we have an increasing population of non-HIV infected infants and children who have received different modalities of immunomodulators (i.e.
, steroids) for various conditions including BMT/SCT, solid-organ transplant and systemic autoimmune/inflammatory disorders that put them at risk for CMV lung infection [7
]. In addition, better molecular diagnostics (i.e.
, CMV qualitative real-time PCR) as well as great progress in performing pediatric bronchoscopy and detailed lung imaging (i.e.
, high-resolution computerized tomography, HRCT) have improved our ability to detect CMV pulmonary infection, re-defining the concept of CMV pneumonia, which is now known to occur in children that are not severely immunocompromised [8
The goal of this article is to characterize CMV lung infection in non-HIV infected children using current molecular and imaging diagnostic modalities, in combination with traditional respiratory signs and symptoms. To this end, we present a case series of 15 children with CMV lung infection focusing on their clinical presentation, radiological patterns, bronchoscopic findings and the molecular approaches used. Our results highlight that the diagnosis of CMV pneumonia in children is challenging but the common clinical and radiological patterns such as hypoxemia, diffuse adventitious lung sounds and ground-glass pulmonary opacities, can provide critical clues to allow early identification of CMV lung infection in the pediatric population.
CMV is a herpes virus that can produce life-threatening pulmonary infections in immunocompromised hosts [4
]. Although effective antiviral therapy for CMV is available [11
], timely diagnosis remains a major challenge for this condition, particularly in the pediatric population where CMV often presents with atypical patterns of lung infection [5
]. The diagnosis of CMV pneumonia can be even more difficult in situations when there is no high-index of suspicion, for instance, in patients that do not have an underlying immunodeficiency syndrome (i.e.
, HIV-infected patients). Indeed, a detailed clinical characterization of CMV lung infection in this age group is critical to select patients that may warrant invasive procedures to obtain lung samples (i.e.
, pediatric bronchoscopy or open lung biopsy) and may benefit from prompt CMV therapy before confirmatory testing is available. Given the paucity of data about the clinical features of CMV pulmonary infection in children, this paper aims to fill this gap in the literature detailing the clinical and radiological features of this condition in a case series (n
= 15) of non-HIV infected pediatric patients with pulmonary CMV infection. Our data illustrates that despite significant variability in the clinical presentation, there are specific features consistently present in pediatric CMV pulmonary infection, including hypoxemia, diffuse adventitious sounds and ground-glass consolidation pattern in lung CT scan, which together may offer crucial clues in the diagnosis of this condition in children.
As previously reported in the adult literature, we found that the majority of children that developed CMV pneumonia were immunocompromised (87%) and 73% of patients were on systemic steroid therapy for different conditions including asthma, systemic lupus erythematous, leukemia treatment, renal transplant and allogeneic bone marrow/stem cell transplant. In addition, 8/15 of the subjects (53%) had hypogammaglobulinemia, which was mostly secondary to steroid treatment and/or malnutrition in our case series. These risk factors have been previously reported in the literature [15
], and reflect the need to have proper humoral and cell-mediated immunity to clear CMV infection in the lungs [3
]. During the study period we did not identify children with HIV and pulmonary CMV in our institution. This may reflect the population treated by the authors and not the overall trend of CMV infections in pediatric HIV. In this regard, it is important to emphasize that children with HIV may have different patterns of CMV lung disease. For instance, HIV itself is often associated with non-malignant lymphocytic infiltrative disorders, including nonspecific interstitial pneumonitis and lymphocytic interstitial pneumonitis (LIP) [18
], which could potentially change the radiological appearance and the clinical manifestations of opportunistic lung infections like CMV. In addition, HIV patients might have more extra-pulmonary manifestations of CMV (retinitis and hepatitis) as well as other opportunistic lung pathogens such as Pneumocystis Jirovecci [19
] that may lead to more severe disease and worse prognosis.
In terms of the clinical presentation and respiratory compromise, the onset of symptoms had a mean time of 14 days however there was considerable variation ranging from three days in a patient that required early mechanical ventilatory support, and up to three weeks in one child with nonspecific symptoms, a feature that has previously labeled CMV as an unpredictable respiratory infection [5
]. Overall there was also great heterogeneity in the severity of the respiratory compromise. While one patient had self-limited clinical course that did not required therapy with ganciclovir, another had severe CMV infection and died despite antiviral therapy. In contrast, there was a consistent homogenous pattern of initial respiratory symptoms/signs that included a combination of cough, hypoxemia, increased breathing effort (i.e.
, retractions or dyspnea) and diffuse abnormalities in lung auscultation, which were present in virtually all cases (Table 3
). The diffuse adventitious lung sounds identified were described as either wheezing, rales and/or rhonchus, which suggests variable degrees of involvement of lung parenchyma and conductive airways during CMV infection, compatible with what is generally described in CMV lung pathology [20
]. In association with respiratory symptoms, about half of the patients had constitutional manifestations (fever or weight loss) and accompanying laboratory abnormalities that included anemia (53%), thrombocytopenia (40%), leukocytosis (40%), leukopenia (27%) and elevated liver enzymes (47%). These abnormal laboratory findings are in overall agreement with prior reports of CMV lung infection in children [11
One of the most important findings of our study was the current difficulty/delay in establishing the diagnosis of pediatric CMV lung infection. There was a median time to diagnosis of 26 days (IQR 11–37 days), which could be explained by the low clinical suspicion in the initial management of these non-HIV infected neonates and children. Indeed, relatively prompt diagnosis was obtained in those individuals with BMT, SMC and bone marrow transplantation (Figure 2
, Figure 4
and Figure 5
), which are well-known risk factors for CMV and therefore bronchoscopy/BAL was performed early. On the other hand individuals without underlying immunodeficiency (i.e.
, steroid-dependent asthma; Figure 6
) or malnutrition were subject to longer time to diagnosis. Another factor that contributed to CMV diagnosis delay was the turnover time for the molecular studies. The CMV molecular diagnostic approach was performed using several modalities, including CMV viral load in peripheral blood, CMV qualitative PCR amplification in bronchoalveolar lavage (BAL) fluid and IgG serology for CMV. Prior reports have been found in which measurements of CMV viral DNA done in samples of BAL or sputum are better to make a diagnosis of CMV pneumonia, compared with the gold standard demonstration of cytomegalic inclusions in lung tissue [22
]. Honda, J., et al.
studied 363 CMV adult patients with 882 samples of sputum, BAL, peripheral blood and urine, showing a positive predictive value (PPV) and negative predictive value (NPV) of 100% and 98.8% for BAL samples, 95.5% and 99.7% for sputum samples; and a sensitivity 90.9% and specificity 100% for BAL and 95.5% and 99.7% for sputum [22
]. In our pediatric case series, CMV PCR in BAL was obtained in 11/15 and was positive in 82% of the cases (9/11); which may reflect the technical difficulties of obtaining a proper BAL specimen in the pediatric population, particularly in neonates [9
]. Interestingly, most children with CMV lung infection had neutrophilia instead of lymphocytosis in BAL cellularity, which has been described in the adult literature [7
]. The later may be attributed to bacterial superinfection, immunosuppression (affecting lymphocyte function and proliferation) or differences in the airway immune response to CMV in the pediatric population. In addition to bronchoscopy, four patients also underwent lung biopsy, two of these had PCR negative for CMV in BAL and the other two had a PCR positive in BAL but biopsy was performed due to additional concerns about other pulmonary infectious/inflammatory processes (i.e.
, persistent hypoxemia and alveolar hemorrhage). There was one patient who did not undergo a flexible bronchoscopy with BAL before open lung biopsy. In all five cases, lung biopsy confirmed the diagnosis of CMV, demonstrating cytopathic changes (Figure 6
) and positive immunohistochemistry (Figure 1
and Figure 3
). Additional pathological findings in lung biopsies were diffuse monocytic infiltration (Figure 1
, Figure 3
and Figure 6
) alveolar hemorrhage and pulmonary vascular disease in a patient with history of hemosiderosis and pulmonary hypertension. Because not all patients had the same CMV testing done (i.e.
, viral load) we could not do correlations of laboratory values with clinical parameters or the severity of CMV lung infection.
The initial imaging diagnostic approach consisted of chest radiography and later chest CT to further characterize the parenchymal involvement and to help in the decision making for further testing such as flexible bronchoscopy and/or lung biopsy. One hundred percent of patients had abnormalities on radiological exams; the most frequent findings were consolidation, usually compromising the dependent lung regions (bi-basal consolidation; Figure 3
) and ground-glass opacities caused by the partial displacement of air due to filling of alveolar spaces, interstitial thickening and/or partial collapse of alveoli leading to enhanced small airway lumen marks, which is a radiological sign known as “dark bronchus” [10
] (Figure 2
). The ground glass/consolidation pattern has been previously identified in adults with CMV infection [24
]. These radiological findings underlie the clinical presentation of CMV lung infection in our pediatric series that included increased respiratory effort, abnormal breath sounds and hypoxemia secondary to diminished alveolo-capillar gas diffusion and abnormal ventilator (V)/ perfusion (Q) match in the lungs [25
]. Importantly, CT scan was superior to CXR in detailing the radiological pattern of CMV lung infection in our pediatric case series. This is in agreement with Smith et al.
who previously reported children with CMV pulmonary disease having minimal non-specific abnormalities in CXR [5
]. In this context, it is noteworthy to mention that newer CT scan lung modalities (high-resolution) are now being proposed to diagnose different pediatric pulmonary conditions and thus avoid open lung biopsies [27
]. For instance, some types of children’s interstitial lung disease (chILD) such as neuroendocrine cell hyperplasia in infancy (NEHI), have specific patterns of disease in CT scan that may be sufficient to make a diagnosis when combined with a specific set of clinical features [27
]. Based on our case series, we propose a group of clinical criteria (Table 5
) based on risk factors (immunosuppression), symptoms/signs and lung imaging that when present must raise concern for CMV pulmonary infection in children and should prompt further confirmatory investigation (i.e.
, bronchoscopy) or empiric therapy depending on the clinical situation.
Key features of CMV lung infection in non-HIV infected children.
Key features of CMV lung infection in non-HIV infected children.
| Systemic steroid use |
| Malnutrition |
| Hypogammaglobulinemia |
| Hematologic malignancy |
| Increased breathing effort |
| Diffuse adventitious lung sounds (i.e., rales, wheezing)|
| Ground-glass opacity/consolidation|
| CMV PCR in BAL|