Metástases pulmonares em crianças : estamos operando desnecessariamente ? Pulmonary metastases in children : are we operating unnecessarily ?

A maioria dos tumores pulmonares em crianças e adolescentes corresponde a metástases, sobretudo de osteossarcomas. Hoje, para a detecção dessas metástases, o exame mais utilizado é a tomografia computadorizada (TC), que, apesar de ser um exame muito superior à radiografia convencional, possui, ainda, algumas limitações. Dentre elas a impossibilidade de diferenciação entre lesões benignas e malignas, principalmente as calcificadas e pequenas1. A literatura médica é escassa sobre o tema, persistindo dúvidas importantes. Apesar de Brader et al.2 terem encontrado uma relação entre tamanho, presença de calcificação e metástases pulmonares para osteossarcomas em crianças e adolescentes, muitos trabalhos não estratificam as crianças dos adultos e agrupam todos os tumores, independente das características próprias de cada faixa etária3-5. Trabalhos sobre esse tema devem ser estimulados, a fim de ajudar o cirurgião pediátrico e o oncologista no atendimento de crianças com tumores malignos. O objetivo deste estudo é determinar, em pacientes pediátricos portadores de neoplasias malignas, as características de nódulos pulmonares identificados à tomografia computadorizada, capazes de diferenciar nódulos benignos de metástases.


INTRODUCTION
T he majority of lung tumors in children and adolescents correspond to metastases, especially from osteosarcomas.Today, for the detection of these metastases, the most commonly used exam is the computed tomography (CT), which, despite being a much better examination than conventional radiography, still has some limitations.Among them, the impossibility of differentiating between benign and malignant nodules, especially calcified and small lesions 1 .The medical literature is scarce on the subject, important doubts persisting.Although Brader et al. 2 have found a relationship between size, presence of calcification and pulmonary metastases of osteosarcomas in children and adolescents, many studies do not stratify the children from adults and group all tumors, regardless of the characteristics of each age group [3][4][5] .
Works on this topic should be encouraged to assist the pediatric surgeon and the oncologist in the care of children with malignant tumors.The objective of this study is to determine the characteristics of pulmonary nodules identified in computed tomography in pediatric patients with malignant neoplasms, as well as the possibility of differentiating benign lesions from metastases.

METHODS
We carried out a retrospective study of pediatric patients undergoing exploratory thoracotomy to approach possible pulmonary metastases in a quaternary hospital over a sevenyear period (April 2009 to April 2016).The preoperative image evaluation was performed with high resolution CT.We excluded patients with primary thoracic tumors from the study.We studied patients with extrathoracic primary tumor submitted to thoracotomy for evaluation of pulmonary metastases, assessing age, surgery performed, primary diagnosis, location, nodule size, presence of lymph node enlargement on CT scan, and anatomopathological examination with or without confirmation of malignity.
We approached the data through descriptive analysis, and calculated the central tendency (median) measures of the largest and smallest tumor diameters for the cases with and without metastasis.We constructed the Receiver Operator Characteristic (ROC) curve with the largest nodule diameters at CT that were subsequently diagnosed as metastasis, to find the best cutoff point for diagnosis.From that point, we calculated sensitivity, specificity, negative and positive predictive values.The programs used for analysis were R version 3.2.5,Stata version 14 and EpiInfo 7 th edition.
The study was approved by the institution's Ethics in Research Committee, protocol numbering 1,425,036.

RESULTS
During the study period, 31 thoracic interventions were performed in 25 patients for resection and/or biopsy of lung lesions suspected of malignancy.Sixteen patients had primary tumor of intrathoracic location and were excluded from the study.Thus, we selected nine patients submitted to 11 interventions for this research.
The mean age of the operated patients was 8.6 years.One patient underwent bilateral thoracotomy at different times and another was reoperated for excision of new nodules from the same previously operated hemithorax.Six of the nine cases had osteosarcoma as primary diagnosis, and the others, nephroblastoma, hepatoblastoma and teratoid tumor.
In four interventions (out of 11), the number of nodules was greater than five, and they were bilateral in five.The size of the possible metastases ranged from 2mm to 75mm.Only one patient had lymph node enlargement at CT.This patient had previous diagnosis of hepatoblastoma and no sign of malignancy was found in the specimen sent to pathological anatomy (Table 1).In only five samples (out of 11), signs of malignancy were confirmed at the anatomopathological examination.In the others, no cellular atypia or mitotic activity was observed.The lesions were diagnosed as foci of hematopoiesis (one case), inconclusive (one case), reactional lymph node (one case), calcified nodules (two cases), and in one case, described as inflammatory nodule (Table 1).For the cases with metastases, the median of the nodules' smallest diameters identified on CT was 6.5mm, and the largest diameter, 25mm.For the cases without metastases, the medians were 3mm (smaller diameters) and 5.5mm (for the larger ones).In the present study, we identified a correlation between the largest nodule diameter found on CT and malignancy.
For this, we constructed a table, ordering all the nodules by their largest diameter, and with each value we calculated sensitivity and specificity using this parameter as cutoff (Table 2).Thus, using the cutoff size of the smallest nodule of 2mm x 2.9mm, to say that the CT node is malignant, we would have a sensitivity of 100% (all malignant tumors would be found), but a specificity of 0 and accuracy of only 45.45%, that is, 45.45% of patients in this series would have the correct diagnosis at CT (Table 2).When we considered the cutoff point of 12.5mm (<12.5mmbenign and ≥12.5mm malignant), we observed a sensitivity of 100%, with a specificity of 83.33% and an accuracy of 90.91%, these being the best values found (Table 2).Finally, was calculated the area under the ROC curve with the data obtained and the p-value of the equality test for statistical validation.We observed that the area under the ROC curve was 0.967, 95% confidence, interval 0.874-1.000,with a p-value of 0.011.Therefore, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) from the cut-off point of 12.5mm were 100%, 83.33%, 83.33%, and 100%, respectively.
Other features described as number of nodules, bilateral lesions or lymph node enlargement did not display statistical significance in differentiating patients with benign nodules from those with malignant ones, probably due to the small sample.

DISCUSSION
Resection of secondary pulmonary nodules remains a part of the curative treatment for most pediatric oncological diseases [4][5][6][7][8] .Cliffton and Pool, in 1967, established the key points for the treatment of pulmonary metastatic disease in children, which are: 1) the number of metastases cannot be a contraindication to surgery; 2) the type of tumor must be respected; 3) staged resections are well tolerated; and 4) when in doubt, surgical excision is to be performed, to give the child a chance of healing 9 .
Although chest CT has issues, such as failure to detect some metastatic nodules, difficulty in differentiating benign and malignant lesions, especially in calcified lesions, and radiation exposure, it is still the recommended examination for staging, with better accuracy for detection of lung nodules than chest X-ray 10 .Other tests, such as magnetic resonance imaging and positron emission tomography, are still in the improvement phase to detect these metastases and are not readily available technologies 11,12 .
Once the pulmonary nodule has been diagnosed, in our Service, all children undergo surgical resection, regardless of nodules' size, in view of the lack of significant data in the literature to differentiate between benign and malignant nodules on chest CT.There is a preference for thoracotomy, due to the possibility of palpation of new lesions missed by the tomographic examination.In the presence of bilateral lesions we opt for a two-time treatment, with posterolateral thoracotomy in one hemithorax followed by thoracotomy in the other hemithorax about 15 days later.These conducts are accepted in the literature, although there is controversy as to the best access route (thoracotomy versus thoracoscopy).Another issue that exists in the preoperative period is the need to stage surgeries for bilateral lesions or to operate them in a single time 10,[13][14][15][16][17][18][19][20] .Thoracoscopy is a promising access route, but it is better indicated in peripheral and small lesions 10,21 .Although there are studies that advocate its use, justified by the good correlation of tomography with intraoperative findings, it is known that CT accuracy is still insufficient 22,23 .The finding of lesions through palpation, not detected by CT, is frequent.We also ought to remember that intraparenchymal lesions offer greater technical difficulty for their excision through thoracoscopy.Thus, we chose to perform thoracotomy, staged when in the presence of bilateral lesions, given the greater pain tolerance of these patients with this type of conduct. It

Table 1 .
Patients, their primary diagnosis, the anatomical-pathological result and the nodules' features at CT.

Table 2 .
Cutoff values of the greater tumor diameter according to sensitivity, specificity and accuracy.