Selection of the appropriate patients for radical local treatment, whether surgery, radiotherapy, or multimodal treatment, is essential. Today, FDG-PET for the staging of nodal and distant metastases is the standard of care, based on international guidelines. In a prospective study of NSCLC patients with mostly locally advanced disease, initial staging was performed using CT and with FDG-PET thereafter. FDG-PET led to a change from curative to palliative therapy by upstaging the disease extent in 25% of the patients [6]. A similar study was performed in a patient cohort referred for radical radiotherapy based on CT staging [7]. The 76 patients were mostly cN+. After FDG-PET staging, radical radio chemotherapy was performed in only 66% of the patients, with palliative treatment in the remaining 34% mostly because of the detection by FDG-PET imaging of distant metastases or of more extensive nodal disease considered too extensive for successful radical radiochemotherapy. At 4 years after treatment with curative and palliative intent, OS was 35.6% and 4.1%, respectively, thus confirming the accurate selection of a high-risk population using FDG-PET staging.

The timely performance of FDG-PET staging, before the start of radical treatment, is crucial. If the interval between staging and the start of radical treatment is too long, FDG-PET needs to be repeated. Everitt et al. evaluated two sequential FDG-PET/CT images acquired from 82 patients with a median interval of 24 days prior to the start of treatment [8]. InterScan disease progression (TNM stage) was detected in 11 (39%) patients. Treatment intent changed from curative to palliative in 8 (29%) because of the detection of newly developed distant metastases in the second FDG-PET/CT; in 7 of these patients the change in treatment was based on the PET image component. The average standardized uptake value (SUV) increased by 16%.

In patients with NSCLC, FDG-PET is especially useful for nodal staging, which is of fundamental importance in therapeutic decision-making (curative vs. palliative; surgical vs. non-surgical; conventionally fractionated radiotherapy vs. stereotactic body radiotherapy) and for target-volume definition in radical radiotherapy. It is well documented that nodal staging using CT imaging is of low sensitivity and specificity. This is especially true in patients with cN0 disease. D’Cunha et al. reported the results of the CALGB 9761 study of 502 patients with clinical stage I disease as determined by CT staging [9]. After surgical resection and mediastinal lymph node dissection, 14% of the patients were shown to have had pathologic stage II disease and another 13.5% stage III disease.

Two studies evaluated the value of FDG-PET staging in patients with clinical stage IA disease based on CT staging. Park et al. reported a retrospective study of 147 patients who had clinical stage IA disease according to FDG-PET staging [10]. After systematic lymph node dissection in the majority of the patients, 14.3% had occult nodal (N1 or N2) metastasis. Total N1 and N2 involvement was detected in 9.5% and 4.8%, respectively. Multivariate analysis demonstrated that a primary tumor with a SUVmax >7.3 was an independent predictor of occult nodal metastasis. In the retrospective study by Stiles et al. of 266 patients with stage IA disease according to CT-and FDG-PET-based staging, mediastinal lymph node dissection detected N1 and N2 disease in 6.8% and 4.9% of the patients, respectively [11]. Tumor size >2 cm and FDG-PET positivity were risk factors for understaging by FDG-PET.

Overall mediastinal staging accuracy was evaluated in a meta-analysis. For CT staging, the median sensitivity and specificity of mediastinal staging were 61% and 79%, respectively. For FDG-PET, the corresponding values were 85% and 90. However, the specificity of FDG-PET staging was lower when CT showed enlarged lymph nodes [12]. Consequently, all positive nodal findings in FDG-PET need to be confirmed pathologically, using endobronchial ultrasound or mediastinoscopy.

In patients with stage I NSCLC based on FDG-PET staging, stereotactic body radiotherapy (SBRT) of the primary tumor only, without elective treatment of the hilar or mediastinal lymph node regions, is the guideline-recommended standard of care in all patients who are medically inoperable [13]. The omission of elective nodal irradiation combined with other high-precision technologies (respiration correlated imaging, image guidance, intra-fractional motion management) allows for irradiation with escalated irradiation doses, which are delivered in a hypo-fractionated manner: biological equivalent doses (BED) >100 Gy are delivered in 1–8 fractions.

Local tumor control in SBRT is significantly improved compared to conventionally fractionated radiotherapy and reaches >90% in the majority of studies. This improved local control translates into significantly improved OS [14], which is mainly limited by the comorbidities of the patients as well as systemic progression of the disease. Regional failures are rare after FDG-PET staging. Senthi et al. reported a 7.8% regional failure rate at 2 years in a large cohort of 676 patients [15].

In locally advanced NSCLC, involved-field irradiation without elective nodal irradiation is currently practiced in most studies and centers, if FDG-PET had been performed for staging purposes. Involved-field irradiation reduces the irradiated volume substantially, thus allowing either a reduction of toxicity or iso-toxic escalation of the irradiation dose [16]. Early studies confirmed low rates of isolated regional failures in un-irradiated hilar and mediastinal regions [17], with only 1 out of 44 patients treated with involved-field irradiation developing an isolated nodal failure. The value of involved-field irradiation after FDG-PET staging is currently being evaluated in a prospective randomized multi-center trial (PET-Plan, NCT00697333).

The correlation of pre-treatment, intra-treatment, or early post-treatment FDG-PET characteristics with outcome could be used for patient stratification and subsequent treatment adaptation. However, whether FDG-uptake is a prognostic maker in NSCLC is discussed controversially in the literature. The uncertainty is at least partially explained by differences in the methodology of outcome modeling. Firstly, different endpoints have been used to determine a correlation with FDG-PET characteristics: local tumor control, progression-free survival, and OS. Secondly, different metrics of FDG-PET images have been used for outcome correlation: SUVmax, metabolic tumor volume, and more recently, texture characteristics such as coarseness, contrast, and busyness. Additionally, tumor volume is a well known independent prognostic marker, which could confound analysis using FDG-PET as a prognostic marker [18]. Another matter of controversy is whether or to what extent inflammatory reactions during or shortly after radio(chemo)therapy influence the value of FDG-PET for outcome modeling.

The prognostic value of pre-treatment FDG-PET was evaluated in a meta-analysis of 21 studies by Paesmans et al. [19]. The endpoint was OS. Data from individual data patients were not available; instead the median SUV value of each study was used as a threshold. The study detected a poor prognosis for patients with a high vs. a low SUV, with an overall combined hazard ratio of 2.08. However, it did not find an optimal SUV threshold but only that higher SUVs resulted in worse outcome. This may have been a consequence of the limitations of the meta-analysis or that rather than two groups of patients there was a continuous increase in the hazard with increasing SUV

Matchay et al. reported one of the largest prospective studies, in which the pre- and post-treatment FDG-PET characteristics of 250 patients with locally advanced NSCLC treated with conventional concurrent platinum-based radiochemotherapy without surgery were analyzed [20]. The 2-year survival rate for the entire population was 42.5%. Neither pre-treatment SUVpeak nor SUVmax correlated with OS, which is in contrast to the metaanalysis described above. In contrast, SUVpeak and SUVmax in FDG-PET images acquired approximately 14 weeks after treatment correlated with OS. Patients with higher residual FDG-PET uptake had a significantly worse OS.

In a further analysis, Aerts et al. analyzed the spatial correlation between pre- and post-treatment SUV in 55 patients treated with chemoradiation for locally advanced NSCLC [21]. Pre- and post-treatment FDG-PET images were acquired about 2 weeks and 12 weeks prior to and after radiotherapy, respectively. The authors reported that patients with residual metabolic-active areas within the tumor had a significantly worse survival than patients with a complete metabolic response. Most importantly, the location of residual metabolic-active areas within the primary tumor after therapy correlated with the initially high FDG uptake areas determined pre-radiotherapy. Consequently, pre-treatment FDG-PET imaging could be used to identify subvolumes of the primary tumor at risk for incomplete response after definitive radiotherapy.

A similar study was performed by a group from the University of Michigan [22]. In 15 stage I–III NSCLC patients treated with a definitive dose of fractionated radiotherapy, pre-treatment (2 weeks), intra-treatment (after approximately 45 Gy), and post-treatment (3–4 months) FDG-PET/CT images were acquired. A significant correlation between metabolic tumor response during radiotherapy and metabolic tumor response 3 months post-radiotherapy was determined.