Educational Aims
The reader will be able to:
- •
Review the main features of necrotising pneumonia in children.
- •
Understand the areas where there is a need for more evidence to guide management.
- •
Realise that long-term evolution is favourable, and deaths are rare.
Abstract
The objective of the review was to determine the long-term outcomes of necrotising pneumonia (NP). Studies published since 1990 in English, Portuguese, or Spanish, published on PubMed and Scielo were evaluated. Our findings showed ultrasound scanning is the diagnostic modality of choice. Despite prolonged hospitalisation (median 13–27 days) and fever (median 9–16 days), most patients recover completely. Empyema and bronchopleural fistulae are frequent in bacterial NP. Streptococcus pneumoniae is the most prevalent cause. Seventeen studies with 497 patients followed for 30 days to 8.75 years showed that most patients were clinically asymptomatic and had normal lung function. X-ray or CT chest imaging demonstrated that almost all lung lesions recovered within 4–6 months. We suggest that it is not necessary to request frequent chest X-rays during the treatment and recovery process. Chest CT scans should be reserved for specific cases not following the expected clinical course.
Introduction
Necrotizing pneumonia (NP) is characterised by the development of necrosis, liquefaction and cavitation of the lung parenchyma by an infectious pathogenic agent in a child with community-acquired pneumonia (CAP). NP is a serious condition and may develop regardless of early and adequate treatment of the initial CAP. .
The incidence of NP is increasing globally and affects about 3.7 % of hospitalised children with CAP and is present in 40 % of complicated pneumonias. It mainly affects children aged 1–4. Diagnosis is based on the combination of clinical signs of pneumonia and imaging and may be confirmed by pathological studies, or occasionally, by intraoperative findings. NP has a high rate of complications during the acute phase, with significant morbidity. However, antibiotic treatment leads to significant recovery in most patients before hospital discharge. In selected severe cases, intervention may be necessary, but this is rare. The aim of this article is to review the long-term outcomes of CAP complicated with lung parenchymal necrosis (NP) in children.
Methodology
We conducted a non-systematic review, which included studies since 1990 that were either complete articles or published abstracts in English, Portuguese, or Spanish, published on PubMed and Scielo, to evaluate long-term outcomes. The search terms used were ‘necrotizing pneumonia’ and ‘children’ and ‘long-term’, ‘follow-up’ or ‘outcome’ in combination, limited to titles and abstracts. After reviewing the abstracts, we included studies that provided information on clinical assessment, lung function, or imaging findings after hospital discharge. Follow-up studies of complicated pneumonia with separate NP analysis were also included. Studies with less than three patients or other causes of pulmonary necrosis or complex chronic conditions were excluded ( Fig. 1 ).
Physiopathology
The detailed mechanism is not well established but appears to be related both to host factors and bacterial virulence. The mechanism may be that infection causes vasculitis, thrombosis of intrapulmonary vessels, causing ischemic damage, liquefaction, and necrosis in the consolidated lung. Drainage of necrotic contents through the bronchial tree results in the formation of multiple cysts or cavities and temporary loss of the normal architecture of the lung parenchyma. .
Clinical manifestations and diagnosis
The clinical signs and symptoms of pneumonia are typically present, and the possibility of necrosis should be considered in all children with pneumonia, particularly in cases where signs of infection persist for 48 to 96 h without improvement despite appropriate antibiotic treatment, empyema drainage, or if there are bronchopleural fistulae (BPF). Another presentation is with clinical signs of severe, fulminant and progressive pneumonia, progressing to septic shock and respiratory failure.
Fever persists for a mean of 9 to 16 days after admission. Patients with NP characteristically experience prolonged fever compared to those with non-necrotizing pneumonia complicated by empyema or pleural effusion (PE). .
Bacterial necrotizing pneumonia is more severe, has more systemic complications such as abdominal pain, diarrhoea and haematogenous spread, is more likely to require Intensive Care Unit (ICU) admission, and is characterised by a longer length of stay (LOS) than Mycoplasma pneumoniae NP. Patients in whom bacteria are isolated tend to have more complications and require more invasive procedures such as mechanical ventilation .
Children infected with Staphylococcus aureus producing Panton-Valentine leukocidin (PVL + ) pneumonia present with more severe disease, with a higher frequency of sepsis/septic shock, admission to ICU requiring mechanical ventilation, and empyema, and a 5-fold higher need for pleural drainage compared to children with pneumococcal pneumonia. There is a higher mortality rate in presentations with airway bleeding, erythroderma, or leukopenia. Coinfection of influenza A and S. aureus- PVL + is associated with severe pneumonia and a higher frequency of complications and mortality. Pulmonary and extra-pulmonary complications are frequently associated, mainly pleural effusion (PE) and empyema. Ness-Cochinwala et al. reported that septic shock occurred in 14 % of cases. Approximately 25 % of patients required mechanical ventilation, which contributed to prolonged hospital stays and increased morbidity.
Lai et al. observed haemolytic uremic syndrome in 10 % of patients, all with positive cultures for Streptococcus pneumoniae . The frequency was significantly higher (20 % vs. 3.2 %) in children with extensive necrosis. .
Mortality ranges from 0 to 7 % but is doubled in patients with complex chronic conditions [2,4,6,7,11,1.4] It increases to 50 % in NP a due to S. aureus PVL +. .
The length of stay (LOS) in the hospital may be prolonged by an average of 13 to 27 days, even when antibiotics are used before hospitalisation, especially in patients with more severe lung involvement. However, children with Mycoplasma NP have an average of twice the LOS than those without necrosis. In patients undergoing surgery (decortication for empyema and debridement of necrotic areas), the LOS can be as long as 27 days. .
Imaging
Chest radiography (CXR), computed tomography (CT), or ultrasound (US) findings confirm the presence and characterise the extent and distribution of areas of necrosis. Imaging may identify extrapulmonary complications, such as PE and bronchopleural fistula [BPF], exclude differential diagnoses and guide management. Although the accuracy of CXR is inferior to CT in the visualisation of necrosis, this is generally the imaging modality that suggests the diagnosis. US has a similar sensitivity to CT for identifying NP, is superior in visualising septations in PE and also helps to guide thoracocentesis. If used with Doppler, the identification of impaired perfusion can predict massive necrotic changes early, even before they are apparent on CT. In addition, US can monitor treatment response and predict outcomes, with the added advantage of not exposing the patient to radiation. An advantage of CT over US is the differential diagnosis of necrosis with loculated hydro- or pyo-pneumothorax. CT should be reserved exclusively for severe cases with an unfavourable evolution, primarily for the differential diagnosis of other complications such as lung abscess and for pre-operative planning if surgery for empyema or BPF is contemplated. It should be performed with intravenous contrast administration to differentiate consolidated areas, pulmonary necrosis, and PE. Low-dose CT scan is an excellent option. Recent studies have attempted to adopt a machine-learning radiomics model based on radiographic features observed in CT scans to recognise lung consolidation in the early stage of NP in children to reduce radiation exposure. In this study, the authors built a model combining regular CTs without contrast with an Artificial intelligence system (radiomics), which selects areas of the lung before being diagnosed as necrotic so could anticipate the emergency of necrosis. This method has been used in other lung pathologies but needs prospective studies to show its utility in NP. .
Some degree of scoliosis can be detected in x-ray in more than 50 % of patients with pleural effusions (mainly in those with late diagnosis). A few weeks later, the scoliosis has resolved without any specific treatment. .
Aetiology
Table 1 presents the most frequent etiological agents identified in this bibliographic search in 22 studies with 1,408 patients. Inclusion criteria, number of patients and diagnostic methods varied greatly between studies.
| Subjects included on studies (*) | Positive blood culture | Positive pleural culture | Positive molecular on blood | Positive molecular on pleural fluid | Not discriminated | Total | |
|---|---|---|---|---|---|---|---|
| S. pneumoniae | 1280 | 18 | 65 | 24 | 78 | 81 | 266 (21 %) |
| S. aureus | 1116 | 3 | 1 | 0 | 4 | 49* | 57 (5 %) |
| Streptococcus Group A | 108 | 0 | 2 | 2 | 8 | (**) | 12 (11 %) |
| M. pneumoniae # | 197 | 0 | 0 | 0 | 5 | 73 | 78* (39 %) |
The causative pathogens identification varied among studies depending on several features such as: method of identification (culture, site of the sample, PCR (molecular) and others). Less than 10 % of blood cultures are positive, probably due to previous use of antibiotics. Sensitivity is higher in pleural fluid culture, but culture-independent methods have a better yield and are a good alternative in the etiological definition of culture-negative cases. Techniques include quantitative polymerase chain reaction (qPCR) and immunochromatographic assay for antigen detection. Perez et al. observed that in 81 % of culture-negative pleural fluid, pneumococcal DNA was detected by qPCR results, suggesting improved sensitivity of qPCR over culture in detecting S. pneumoniae , particularly in the setting of preceding antibiotic use. They also observed positive results for pneumococcal antigen detection in 63 % of the samples. The antigen detection test is a simple bedside test that produces results in 15 min, is inexpensive and can be a good alternative, especially in settings with limited resources . Another study identified pneumococcal antigen in 67.6 % and pneumococcal genes by qPCR in 70.6 % of pleural fluid samples, showing a moderate agreement (K = 0.518) between the two methods (p < 0.005) . Therefore, ideally, both culture-based and molecular detection techniques should be used to determine aetiology. Unfortunately, in these studies the findings are not specific for NP but only pneumonia with pleural effusions.
S. pneumoniae is the most prevalent, with an identification rate between 18 and 83 %, followed by S. aureus (10 %). In Italy , S. pneumoniae serotype 3 is the most associated with NP occurring even in children with a complete vaccination schedule with PCV13. .
There have been reports of an increase in the prevalence of group A streptococcus . This pathogen is the second most common cause of infection in some studies. In addition, Haemophilus influenzae was found in patients even though they had been vaccinated against HiB. Additionally, the frequency and morbidity of community-acquired methicillin-resistant S. aureus (CA-MRSA) has significantly increased in the last decade. Although rare, pneumonia caused by S. aureus producing Panton-Valentine leukocidin is potentially serious, with a high mortality rate. PVL is a virulence factor produced by up to 74–100 % of CA-MRSA in methicillin-sensitive (MSSA) and methicillin-resistant (MRSA).
There is an increase in reports of NP due to Mycoplasma , mainly in China. This finding contrasts with other studies where Mycoplasma was either not detected as the cause or was found only in a small percentage of cases (2.9 %). Co-infection with viruses, bacteria, or Mycoplasma has been described. .
Risk factors and predictors of complications
Nine studies evaluated risk factors for necrosis in children with pneumonia. Pre-hospital treatment with ibuprofen is associated with a 2.5 times higher odds ratio for CAP complications, as well as pulmonary necrosis [OR 2.54 CI (1.31 – 4.94); p = 0.008). There is a dose effect with higher cumulative doses of ibuprofen being associated with an increased rate of CAP complications . Some studies have identified factors predictive for pulmonary necrosis, including duration of fever, namely WBC (≥15.1 × 109/L) and CRP (≥121.5 mg/L) . In one study, the duration of fever was longer in NP (18 days NP vs . 11 non-NP) and in another study, it was 9 days NP vs 7 in non-NP. Independent risk factors for identifying a higher degree of pulmonary necrosis include CRP, reduced albumin, D-dimers, and IgM cut off of 122 mg/L, 30.8 g/L, 1367.5 ng/mL, and 95.7 mg/dl, respectively. Additionally, predictive models that incorporate clinical and biological markers have been developed to identify and assess the severity of PN early on. However, studies using these models are restricted to Mycoplasma infection and come from China, which may limit generalizability. Even with these limitations, these studies can contribute to the prediction of the presence of NP, but further research on other predictive models are needed.
Management
This is challenging because there are no guidelines, and in many cases, the aetiologic agent is not identified. Treatment decisions have to be taken on a case-by-case basis. The choice of antibiotic is initially empirical and should be based on epidemiology and local clinical guidelines for pneumonia and adjusted according to the results of microbiological tests as soon as possible. Most patients have a high rate of resolution after conservative clinical treatment with high doses of intravenous antibiotics. If bacteria have not been identified, antibiotic therapy should cover pneumococci and staphylococci, and amoxycillin + clavulanic acid would be a good choice, but does depend on local data. .
When NP is caused by S. aureus PVL+, early aggressive empiric antibiotic therapy with a toxin-suppressing agent is essential. The combination of a beta-lactam with a protein synthesis inhibiting antibiotics such as clindamycin or linezolid eradicates the bacteria and reduces the production and effects of the toxin. Vancomycin, used for MRSA, does not reduce toxin production. Draining purulent secretions reduces the burden of toxin, and administration of human immunoglobulin can reduce the cytotoxic effect. .
Retrospective studies report that antibiotics are often administered for prolonged periods, ranging from 21 to 42 days, and a mean of four antibiotics are used per patient. The average length for intravenous antibiotics ranges from 14 to 19 days, while oral antibiotics are generally prescribed for around 21 days after intravenous therapy is stopped. The duration of oral antibiotics used in patients with NP is longer than in patients with pneumonia without necrosis (21 days vs. 18 days for empyema and 10 days for parapneumonic PE; p = 0.002). .
The surgical treatment of NP remains controversial, with the majority of the cases not requiring operative intervention. Uncomplicated NP can be managed conservatively. The majority of cases requiring surgery are those needing pleural intervention due to PE/empyema associated with NP. Video-assisted thoracoscopic surgery (VATS) has been proposed as primary therapy since the use of fibrinolytics is contra-indicated because it can favour air leakage from necrotic peripheral areas of the lung. .
After optimised clinical management, a surgical approach to necrotic areas may be necessary in cases of extensive necrosis with high-output BPF or large pneumatoceles that did not reduce during follow-up. Most studies showed that improvement occurs without surgical intervention. In a USA study (Kids Inpatient Database) with more than 2 million children with pneumonia, diagnosis based on ICD 10th ed (746 had necrosis and excision of lung tissue was performed in only 3.6 %. Other studies have shown that lung resection is performed in 13 to 77 % of children with necrotising pneumonia, with lobectomy being the most commonly performed procedure, followed by wedge resection and segmentectomy. These findings vary according to the surgeon’s experience, and there is currently a tendency towards smaller resections with greater preservation of the lung parenchyma. However, in our experience, the overwhelming majority of patients recover completely without the need for lung resection.
The approach to the pleural space in patients with NP is also variable. Between 46 % and 96 % of patients required pleural drainage in these studies. In a study of 46 patients who required chest tubes, only 6.1 % underwent VATS.
Recently endobronchial treatment of bronchopleural fistula has been proposed as an option. It is relatively non-invasive with some advantages, including preserving lung parenchyma. In a retrospective study with 6 patients, a specific device was successfully used. By using an endobronchial approach in patients with large pneumatoceles, lung resections were avoided. Further studies are needed on endobronchial treatment which could potentially reduce morbidity both in the acute phase and in long-term evolution. .
LONG-TERM FOLLOW-UP
The results of 17 studies with 497 patients followed for 30 days to 8.75 years are presented.
Clinical outcomes
Clinical outcomes were evaluated in 192 children in 4 studies. A study including 32 patients with NP found chest asymmetry (31 %) and a decrease in respiratory sounds (45 %) one month after hospitalisation. However, by 6 months, these signs had disappeared. Similar findings were described by Sawicki et al., who reported 8 readmissions and 3 small pneumothoraces after discharge in 80 children with NP, but symptoms resolved within 2 months. Bover-Bauza et al. evaluated 24 children with a median of 8.75 years after hospitalisation for NP using the standardised ISAAC (International Study of Asthma and Allergies in Childhood ) questionnaire and physical examination. The authors observed that 3/24 had sporadic episodes of wheezing after NP, and 2 had sporadic wheezing in the preceding year. 41.7 % of children had a family history of atopy, 4 children had wheezing before NP, and 20 % were exposed to passive smoking. Three children presented with new uncomplicated pneumonia but did not require hospitalisation. None of the children had symptoms during exercise, and all had a normal physical examination. In another study, after at least 3 years of follow-up, all 56 who underwent surgery showed growth and development within normal limits. .
Despite the severe acute illness, patients tend to have a good outcome. Changes observed during physical examination usually disappear within six months and do not have any long-term effects on growth, development, or exercise capacity. Even if a new pneumonia occurs, it does not necessarily mean that the condition will become more severe. .
Imaging findings
Fifteen studies evaluated imaging findings at follow-up through CXR or CT, with variable numbers of patients, follow-up times and significant losses in follow-up ( Fig. 2 ). .
