Radiologia Brasileira - Publicação Científica Oficial do Colégio Brasileiro de Radiologia

INTRODUCTION

A solitary pulmonary nodule (SPN) is a round or oval pulmonary opacity of up to 3 cm in diameter, surrounded by normal lung parenchyma, that is not accompanied by pleural effusion, pneumonia, or adenopathy⁽¹⁾. A SPN is found in 0.09-2% of all chest X-rays⁽²⁾ and in up to 51% of chest tomography scans in populations at high risk for lung cancer⁽³⁾. The etiological investigation of a SPN depends on its pre-test probability of malignancy^(1,4,5). Cases with low probability of malignancy (< 5%) are usually managed by active surveillance, whereas those with high probability (> 65%) biopsy or surgery is indicated^(1,5). Cases with intermediate probability of malignancy are usually biopsied for diagnostic clarification⁽⁶⁾. However, biopsy is an invasive procedure that can lead to a significant number of false-negative results⁽⁷⁾. The use of PET/CT with ¹⁸F-FDG (¹⁸F-FDG-PET/CT) can help to define the management of intermediate cases by reclassifying them as having low, intermediate, or high probability of malignancy⁽¹⁾, thus reducing the number of invasive procedures and the total cost of treatment⁽⁸⁾. However, there are no studies evaluating the frequency with which ¹⁸F-FDG-PET/CT determines a change in the probability of malignancy classification of SPNs, from intermediate to low or high.

Because ¹⁸F-FDG is a glucose analogue and tumor cells usually have an increased glycolytic metabolism-due to an increased glycolytic enzyme activity and to the overexpression of glucose transporters⁽⁹⁾-the tumor cells have enhanced ¹⁸F-FDG uptake, which makes them visible on ¹⁸F-FDG-PET/CT images. Therefore, it is possible to distinguish benign from malignant lesions in various conditions, including SPNs. Although the uptake in malignant lesions is usually more intense, the distinction is not always clear, because some benign conditions can also present increased ¹⁸F-FDG uptake, leading to false-positive results, whereas some malignant lesions may not have affinity for ¹⁸F-FDG⁽¹⁰⁾.

The SPN pre-test probability of malignancy is defined on the basis of clinical and radiological data. The first model, proposed by Swensen et al.⁽¹¹⁾, is known as the Mayo Clinic model. That model includes the following clinical variables: age; previous or current smoking; history of extrathoracic cancer; nodule size; presence of nodule spiculation; and nodule location. In 2005, Herder et al.⁽¹²⁾ described a new method for calculating the pre-test probability of malignancy. Their model includes all variables of the Mayo Clinic model plus an assessment of ¹⁸F-FDG uptake by the SPN, as seen on ¹⁸F-FDG-PET/CT. This new approach resulted in higher accuracy levels⁽¹³⁾. However, ¹⁸F-FDG-PET/CT is performed in cases classified by the Mayo Clinic model as intermediate probability only⁽¹⁾. Therefore, ¹⁸F-FDG-PET/CT is expected to reclassify intermediate probability cases as low (< 5%) or high (> 65%) probability, probabilistically defining the best management for each case.

The purpose of this study was to determine how often ¹⁸F-FDG-PET/CT changes the SPN pre-test malignancy classification, based on the probability models described, during the clinical decision-making process.


MATERIALS AND METHODS

This was a retrospective study based on the analysis of all ¹⁸F-FDG-PET/CT scans performed between May 1, 2016 and May 31, 2017 at a nuclear medicine center. All ¹⁸F-FDG-PET/CT scans that were requested for SPN investigation were included. The retrospective analysis of the data was approved by the local research ethics committee (Reference no. 76305317.4.0000.5199).

All ¹⁸F-FDG-PET/CT scans were performed with the same device (Biograph 16; Siemens Healthcare, PA, USA) approximately 60 min after intravenous administration of 3.7-4.8 MBq/kg of ¹⁸F-FDG. The PET images were obtained from the base of the skull to the proximal third of the lower limbs in three-dimensional mode, each body segment position being scanned for 3 min. The images obtained were processed by iterative reconstruction (two iterations of eight subsets with a Gaussian filter). Computed tomography (CT) image acquisition parameters included a slice thickness of 5 mm, a voltage of 120 kV, and no intravenous contrast administration. In addition, a high resolution chest CT scan was performed during a maximal inspiratory breath hold in all patients.

Patient medical records were reviewed, and the following data were collected: gender; age; previous or current smoking; history of extrathoracic cancer; results of the SPN biopsy; and results of follow-up chest CT (cases in which a nodule had been stable for two years or had disappeared were considered negative). The images acquired in the dedicated chest CT performed as part of the ¹⁸F-FDG-PET/CT were then analyzed to determine the diameter of the nodule and whether or not there was spiculation. The PET and CT images were fused and evaluated. The nodule uptake of ¹⁸F-FDG was classified, by its maximum standardized uptake value (SUVmax), as follows⁽¹³⁾: discrete, when the SUVmax was ≤ 2.5; moderate, when the SUVmax was 2.6-10; or intense, when the SUVmax was > 10.

Calculation of the pre-test probability of malignancy according to the Mayo Clinic model

The Mayo Clinic model determines the SPN probability of malignancy with the following formula⁽¹¹⁾: PM = 1 / 1 + e − x , where PM is the probability of malignancy and x = −6.8272 + 0.0391 × (age in years) + 0.7917 × (previous or current smoking) + 1.3388 × (history of extrathoracic cancer) + 0.1274 × (SPN diameter in mm) + 1.0407 × (presence of spiculation) + 0.7838 × (SPN located in the upper lobe). Previous or current smoking, history of extrathoracic cancer, presence of spiculation, and SPN location in the upper lobe are taken as dichotomous variables, meaning that when they are absent they get a score of zero and when they are present they get a score of one.

Calculation of the pre-test probability of malignancy according to the Herder model

The Herder model determines the probability of malignancy by making use of the Mayo Clinic model together with the ¹⁸F-FDG-PET/CT results, according to the following formula⁽¹²⁾: PM = 1 / 1 + e − x , where PM is the probability of malignancy and x = −4.739 + 3.691 × (Mayo Clinic probability) + 2.322 × (discrete nodule uptake) + 4.617 × (moderate nodule uptake) + 4.771 × (intense nodule uptake). The presence of discrete, moderate, and intense uptake are treated as dichotomous variables, meaning that if it is absent it gets a score of zero and if it is present it gets a score of one.

Ability of ¹⁸F-FDG-PET/CT to identify high or low probability in cases originally classified as having intermediate probability

Nodules classified by the Mayo Clinic model as having an intermediate probability of malignancy were included in this analysis. In those cases, the probability of malignancy was reclassified according to the Herder model as low (< 5%), intermediate (5-65%), or high (> 65%). ¹⁸F-FDG-PET/CT was considered a determinant of the best course of action in the cases reclassified as having a low or high probability, although not in the cases that were not reclassified.

Statistical analysis

Continuous variables are expressed as mean and standard deviation, whereas categorical variables are expressed as absolute and relative frequencies. To assess the ability of ¹⁸F-FDG-PET/CT to inform practice, the cases were divided into inconclusive (intermediate probability of malignancy) or conclusive (low or high probability of malignancy). We used McNemar’s test to compare the proportions in each model. An additional test was performed only in the cases classified as intermediate probability by the Mayo Clinic model. We used confidence interval (CI) to describe the results of this analysis and considered a p < 0.05 as significant. MedCalc Software, version 18.2.1 (MedCalc Software bvba, Ostend, Belgium) was used in the analyses.


RESULTS

A total of 33 patients was enrolled in this study. The mean age was 68.63 ± 12.20 years, and 17 (51.5%) of the patients were male. The mean probability of malignancy was 43.2 ± 25.6% with the Mayo Clinic model and 47.1% ± 38.9% with the Herder model. The other variables considered in this study are detailed in Table 1.




According to the Mayo Clinic model, the probability of malignancy was intermediate in 23 cases (69.7%), low in 2 (0.6%), and high in 8 (2.4%). When the Herder model was applied in those 23 nodules, 6 (26.1%) were reclassified as low and 8 (34.8%) were reclassified as high, whereas 9 (39.1%) were not reclassified (Figure 1). The difference between the Mayo Clinic model and Herder model, in terms of the proportion of cases converted from inconclusive to conclusive, was 36.4% (95% CI: 16.1-56.6%; p < 0.01), in favor of the latter model. Of the cases that were classified as having a high or low probability of malignancy by the Mayo Clinic model (Figure 1), none were upgraded and only two were downgraded (from high to intermediate probability) by the Herder model. In addition, when only the cases classified by the Mayo Clinic model as intermediate were considered, that difference increased to 60.9% (95% CI: 40.9-80.8%; p < 0.05), and ¹⁸F-FDG-PET/CT was the determinant of the course of action in 14 cases (60.1%). Figure 2 shows cases that were reclassified on the basis of the ¹⁸F-FDG-PET/CT findings.



Of the 33 patients included, nine underwent biopsy for histological confirmation, the probability of malignancy having been classified by the Herder model as high in six and as intermediate in three. All of the high-probability cases were malignant, whereas there were benign and malignant findings among the intermediate-probability cases (Table 2). Another five patients underwent a follow-up chest CT examination at our center two years later. In each of the five, the nodule was stable, had shrunk, or had disappeared.