All patients with suspected lung cancer were initially evaluated by a pulmonologist, who decided the diagnostic examinations.

The workup included a conventional staging based on review of the medical history, a physical examination, the results of a chest and upper abdomen CT scan, bronchoscopy and/or CT-guided biopsy. As established in the international guidelines,2,10,11 a FDG-PET/CT scan was performed in all patients considered candidates for curative treatment (without obvious metastatic disease on CT and without comorbidities contraindicating radical treatment). The FDG-PET/CT scans were performed with a full body PET-CT scanner (General Electric Discovery ST: 47 planes, 3.27mm slice thickness, 128×128 voxels, 1.95mm×1.95mm). CT acquisition parameters were 140kV, 80mA, slice 0.8/2.5mm. Reconstruction was performed using a filtered backprojection algorithm, with FWHM ramp filter of 5.45mm. Patients fasted for at least 6h before the scan and serum glucose concentration was less than 140mg/dl in all patients. PET scanning was started 60min after intravenous administration of 4.62MBq (0.125mCi)/kg of FDG.

All cases were then evaluated by a multidisciplinary team of pulmonologists, thoracic radiologists, nuclear medicine specialists, pathologists, thoracic surgeons, radiation oncologists and clinical oncologists, in order to make therapeutic decisions.

Mediastinal invasion was suspected on CT scan when enlarged lymph nodes with a short axis of more than 10mm were observed. All PET-CT scans were interpreted by visual and semiquantitative analysis. In visual analysis, PET-CT was considered positive if lesion FDG uptake was greater than mediastinal background uptake. Semiquantitative analysis was based on the standardized uptake value (SUV) in a region of interest. SUVs above 2.5 in the lesion area were considered positive. When no correlation between SUV and visual analysis was found, the clinical context of the patient was taken into account to reach a final decision regarding any positive or negative lesion.

If a mediastinal involvement was suspected on CT or FDG-PET/CT, a biopsy of mediastinal lymph nodes was obtained by endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA). If EBUS-TBNA was found negative or not evaluable (without representation of lymphoid cells after at least three needle passes),12 a mediastinoscopy was performed. Patients with confirmed negative mediastinal lymph nodes were treated with surgery, which included a systematic dissection of the ipsilateral hilar and mediastinal lymph nodes.

Demographic, clinical and pathological data of each patient were obtained from the medical history records and included in a database. The clinical stage before and after obtention of FDG-PET/CT scan was defined according to the seventh edition of the cancer staging manual of the American Joint Committee on Cancer.13 Details on changes in treatment decisions as a result of information provided by FDG-PET/CT were also collected.

Clinical assessments of disease in mediastinal lymph nodes obtained by using CT and FDG-PET/CT were classified into binary classes of true-positive, false-positive, true-negative and false-negative (FN) using pathologic evaluation as the reference standard. Sensitivity, specificity and positive and negative predictive values for CT and FDG-PET/CT were calculated. In patients with available mediastinal evaluation by EBUS-TBNA or systematic lymph node dissection in surgery, surgical evaluation was considered the reference standard. Patients who had received chemotherapy or radiotherapy prior to FDG-PET/CT or pathologic evaluation were excluded from the analysis. Statistical analyses were performed using a specialized software (SPSS for Windows v.19).

Between January 2008 and December 2010, 522 patients were diagnosed with NSCLC in our hospital, and FDG-PET/CT was performed in 246 (47.1%). Patient characteristics are summarized in Table 1. The vast majority were men and the predominant histology was adenocarcinoma. Surgery was the most common initial treatment, followed by palliative chemotherapy, concurrent chemoradiotherapy, and radiotherapy alone.

With conventional staging by CT, 87 patients (35.5%) were classified in stage I, 21 (8.5%) in stage II, 90 (36.5%) in stage III and 48 (19.5%) in stage IV. Staging after FDG-PET/CT scan was as follows: six patients (2.4%) with no FDG uptake were classified in stage 0, 90 (36.5%) in stage I, 25 (10.1%) in stage II, 65 (26.5%) in stage III and 60 (24.5%) in stage IV.

FDG-PET/CT did not modify staging in 65.4% of cases. Forty of the 85 patients (34.6%) with stage migration were classified in a more severe stage and 45 in a less severe stage according to FDG-PET/CT data. FDG-PET/CT detected metastatic disease unidentified in conventional staging in 25 patients (10.1%). Of these metastases, four were histologically confirmed. In cases without histological confirmation, the clinical progression of these lesions during follow-up confirmed their malignant nature. FDG-PET/CT also identified concomitant malignancies in other organs (pancreas, rectum and pharynx) in three cases, all histologically confirmed. A detailed presentation of conventional CT and FDG-PET/CT staging is shown in Table 2.

Given the high negative predictive value of FDG-PET/CT in mediastinum evaluation, in those cases classified as N0 by FDG-PET/CT, a tissue sample of mediastinal lymph nodes was not systematically obtained to confirm N0 before surgery. In 13 of the 45 patients in whom stage was reduced after FDG-PET/CT examination, histological confirmation of mediastinum was not available because surgical intervention was not carried out. The reasons were poor respiratory function in eight cases, rapid clinical deterioration in two, treatment rejection by two patients, and diagnosis of simultaneous pancreatic cancer in one case. In the other 32 cases, histological confirmation of mediastinal lymph nodes was obtained by systematic dissection of ipsilateral hilar and mediastinal lymph nodes during surgery.

In 60 (24.4% of all FDG-PET/CT performed) of the 85 patients in whom FDG-PET/CT results modified the clinical stage, stage migration involved a modification of the therapeutic approach. This therapeutic change followed an increase in severity stage in 31 cases (12.6%). FDG-PET/CT indicated a more severe stage in 13 cases that had been classified as stage I or II by CT, and initially candidates for radical surgery. The new classification was stage III, with mediastinal involvement in nine cases (candidates for radical chemoradiotherapy), and stage IV in 4 cases (candidates for palliative chemotherapy). The remaining 18 cases were initially classified stage III by CT, theoretically candidates for radical chemoradiotherapy, and FDG-PET/CT indicated stage IV. As a result of these stage modifications, the use of FDG-PET/CT prevented futile thoracotomy in 13 cases (5.2%) and futile chemoradiotherapy treatment in 18 (7.3%).

FDG-PET/CT results modified the therapeutic approach in 29 patients (11.7%) in whom stage level was reduced, which allowed curative treatment in 26 (10.5%). Seventeen patients (6.9%) were treated with surgery including a systematic dissection of the ipsilateral hilar and mediastinal lymph nodes (13 lobectomies, 3 bilobectomies and one segmentectomy). Six patients (2.4%) received radical radiotherapy (surgery was ruled out due to comorbidity), and three (1.2%) radical chemoradiotherapy (two cases classified stage III by FDG-PET/CT and one treated with induction chemoradiotherapy). Curative treatment was discarded in the remaining three patients due to severe comorbidity. Radical surgery was performed in the 17 cases in which FDG-PET/CT reduced the stage, and clinical and pathological stagings were compared. Staging remained unchanged in 10 cases, and three cases classified as stage I by FDG-PET/CT were considered to be in a more severe stage after pathological examination. Of these, two patients were diagnosed as stage II and one as stage III. Two cases clinically classified in stage II had pathologic stages I and III. One patient classified in stage I was found to be stage III (FN of FDG-PET/CT) and, finally, a case initially classified in stage III had a pathologic stage II (mediastinal invasion was assessed by mediastinoscopy before surgery due to discrepancies between CT and FDG-PET/CT results).

CT results initially classified most patients in our series as stage III. We decided to specifically analyze the impact of FDG-PET/CT in this subgroup of patients. FDG-PET/CT modified the therapeutic approach in 38 (42%) of the 90 patients (36.5%) classified as stage III by CT, and initially candidates for treatment with radical chemoradiotherapy (±surgery). Stage was more severe in 19 patients after FDG-PET/CT, who were therefore treated with palliative chemotherapy. The stage of the remaining 19 was milder, and they were treated with surgery. Details on the T and N classification according to FDG-PET/CT results in patients considered to be in stage III by CT are summarized in Tables 3 and 4, respectively.

Of the 246 cases evaluated with FDG-PET/CT, pathologic evaluation of mediastinal lymph nodes was available in 133 (54%); systematic surgical dissection of the lymph nodes has been performed in 88, and EBUS-TBNA in 45. Given the high negative predictive value of FDG-PET/CT in the evaluation of mediastinal invasion in cases classified as N0, a tissue sample of mediastinal lymph nodes was not systematically obtained before surgery. In 13 of the 45 patients in whom stage was reduced after FDG-PET/CT examination, histological confirmation of mediastinum was not available because the surgical intervention was not carried out. The reasons were deterioration of respiratory function in 8 cases, rapid clinical deterioration of two patients, treatment rejection by two patients, and diagnosis of simultaneous pancreatic cancer in one case. In the other 32 cases, histological confirmation of mediastinal lymph nodes was obtained by systematic dissection of ipsilateral hilar and mediastinal lymph nodes. Mediastinal involvement was not explored in all stage IV cases because the therapeutic strategy (i.e. palliative chemotherapy) had not been modified by its results.

Using pathologic evaluation as the reference standard, sensitivity was 0.57 and 0.68 for CT and FDG-PET/CT, respectively. Specificity was 0.64 for CT and 0.86 for FDG-PET/CT. Positive predictive value was 0.48 for CT and 0.75 for FDG-PET/CT, and negative predictive value was 0.72 for CT and 0.81 for FDG-PET/CT. Statistical analysis results are summarized in Table 5.