Introduction

In 2012, lung cancer is still the most prevailing cancer worldwide, with an estimated 226,000 new cases and 160,000 deaths every year [1]. The prognosis of non-small lung cancer (NSCLC) remains poor despite recent improvements in diagnosis and treatment. As a molecular and quantitative imaging technique, positron emission tomography (PET)/computed tomography (CT) has been proven useful, especially in diagnosis and staging of lung cancer [2]. The standardized uptake value (SUV) is a quantified index of FDG uptake. Several studies have shown that FDG uptake reflects the metabolic activity of the patients with primary NSCLC which can be a prognostic factor [3]. The maximum standardized uptake value (SUVmax) was assigned to an abnormality in the primary tumor with the highest SUV. SUVmax could largely represent the malignant behavior of the tumor with a higher reproductivity; thus, SUVmax is more commonly used in current clinical practice. FDG-PET has been established as a functional imaging technique before treatment which plays an important role in predicting the treatment outcome for NSCLC patients.

In order to provide more information for clinical practice to predict the outcome of NSCLC, we analyzed the possible risk factors affecting overall survival (OS) and progression-free survival (PFS) and determined whether the quantitative degree of FDG uptake by the primary tumor is an independent predictor for the prognosis in NSCLC. This approach might also represent an independent factor to evaluate and predict outcome of patients with NSCLC.

Materials and methods

Patient enrollment

We retrospectively collected a total of 50 patients with non-small cell lung cancer in the study between June 2007 and June 2010. A whole-body scan of FDG-PET/CT was performed before treatment for all patients. All patients received neither chemotherapy nor radiation prior to FDG-PET/CT scan. The patients’ characteristics are shown in Table 1. The median age of the patients was 65 years old (age range 45–87 years old); 31 were male and 19 were female. As for histologic types, 25 were adenocarcinoma cases and 25 were squamous cell lung carcinoma cases. The median tumor length was 23 mm (range 5–79 mm). The 7th edition of the UICC (International Union Against Cancer) staging system was used for staging [4]. Written informed consents were obtained from all patients enrolled.

Table 1 Patient characteristics
Full size table

FDG-PET image acquisition

All patients fasted for at least 4 h before PET/CT examination. After confirmation of the serum glucose level (<1.4 g/L), patients were administered with an intravenous injection of FDG at 3.7–4.0 MBq/kg and then rested about 60 min before undergoing the body scan, as recommended for tumor PET imaging [5].

All patients underwent a PET/CT scan through the Hi-Rez Biograph 16 (Siemens Medical Solutions, Germany), which consisted of a high-resolution three-dimensional LSO PET scanner and a 16-row CT scanner. Based on our study protocols, the patients took a whole-body PET/CT scan from the skull base to the mid-thigh level. Low-dose CT (tube voltage 120 kV; effective tube current 30–250 mA; per tube rotation 0.5 s) was performed for attenuation correction and identifying the tumor’s precise anatomical location before PET acquisition. Immediately after CT, PET was performed in an identical transverse field of view with a section thickness of 5 mm and using 3 min per table position and three-dimensional acquisitions. Patients were allowed to take normal shallow respiration during the image acquisition procedure.

FDG-PET image analysis

PET images were reconstructed using an ICS Navigator with four iterations, eight subsets and a Gaussian filter (4.8 mm). A single-scatter simulation model was used for scatter correction [6]. Attenuation correction was performed based on the CT data. No post-reconstruction smoothing filter was used. The reconstructed spatial resolution was 5 mm in the center of the field of view.

For determination of the SUV, a cylindrical region of interest (ROI) was manually placed over the tumor site on the hottest trans-axial slice. The ROI for each SUV calculation was manually drawn around a focal increased FDG as small as possible on the transaxial image. The maximum SUV (SUVmax) within the ROI was used as the reference measurement. SUVs were calculated from the counts-per-pixel and normalized to body weight:

$${\text{SUV = tissue activity (Bq/ml)/[injected activity (Bq)/weight (mg)]}}$$

where injected activity was decay-corrected from the delay between injection and image acquisition.

Treatment

In our study, pursuant to NCCN Guidelines, for those patients who were rendered inoperable by personal factors or iatrogenic factors, precise radiotherapy was adopted alone for early-stage NSCLC. The total dose of the primary tumor ranged from 60 to 70 Gy (median 65 Gy) in stage I. We adopted synchronized chemoradiotherapy to treat stages II–III NSCLC. For synchronized chemotherapy regimen, radiotherapy and chemotherapy started at the same time: platinum-based doublet chemotherapy was adopted once every 3 weeks for 4–6 cycles; a total of 45–60 Gy was administered in radiotherapy based on patients’ general conditions in stages II–III NSCLC. For the stage IV lung cancer patients with PS scores exceeding 2, supportive treatment was used for palliative care. The platinum-based doublet first-line chemotherapy was given with PS scores below 2. We adopted single agent chemotherapy as PS equal to 2. Patients were treated with chemotherapy which was composed of 4–6 cycles. For some of the stage IV patients, radiotherapy was used for palliative care against primary focus or metastatic focus, for example, partial pain killing effect, etc., and radiotherapy dose was 30–45 Gy.

The regimen of all patients referred to NCCN guide, first-line chemotherapy regimens were implemented without second-line chemotherapy regimens, and all patients did not use targeted drug in order for more comparability among experimental subjects. The specific implementation was assessed by the patient’s performance status, lung function status and cardiac status. The follow-up time was measured from the date of ending the treatment. The observation end point is OS.

Statistical analysis

The SUVmax of the primary tumor was compared with the descriptive statistics, and the correlations between pairs of variables were evaluated by using Spearman’s rank correlation coefficient. The ROC curve was generated to assess the primary tumor SUVmax for lymph nodal metastases. The SUVmax values were divided into two groups using the threshold of primary tumor SUVmax chosen by the receiver operating characteristic (ROC) curve. The OS and PFS values were compared between the two groups using the Kaplan–Meier curve. The comparison of SUVmax with patients’ characteristics before treatment was evaluated by T test. To assess the effect of the patients’ characteristics and other prognostic factors on the outcome, we estimated the correlation between the SUVmax in the FDG-PET of the primary tumor, other prognostic factors (age, PS, UICC stage, histologic differentiation, and pathology), and OS and PFS using univariate and multivariate Cox regression analyses. The OS, PFS, and OS of patients with UICC stages I–IV were calculated by the Kaplan–Meier method from the date of initiation of treatment. All analyses were carried out using the SPSS 11.0. Statistical significance was defined as P < 0.05.

Results

The relations between SUVmax and other prognostic factors (age, tumor length, pathology, T stage, N stage, UICC stage, histologic differentiation) were shown in Table 2 and the correlation coefficients were obtained. Positive correlations were found between SUVmax and N stage (r = 0.855, P < 0.001), SUVmax and histologic differentiation (r = 0.435, P < 0.05), SUVmax and UICC stage (r = 0.345, P < 0.05) before treatment. For the primary tumors, no significant correlations were observed between SUVmax and age (r = 0.314, P = 0.24), SUVmax and tumor length (r = 0.298, P = 0.012), SUVmax and pathology (r = 0.173, P = 0.229), SUVmax and T stage (r = 0.105, P = 0.469).

Table 2 The correlations between patient characteristics and the SUVmax
Full size table

The median SUVmax in the FDG-PET images of the primary tumors was 9.82. The comparison of SUVmax with various patient characteristics before treatment was shown in Table 3. There was a significant difference between SUVmax of the primary tumor and N stage, UICC stage, and histologic differentiation (P = 0.000, 0.000, 0.003). We then divided the 50 patients into two groups according to ROC analysis of their SUVmax values: <5.45 (low-SUV) and ≥5.45 (high-SUV). The description of patient characteristics in the two groups (SUVmax <5.45 and ≥5.45) was distributed in Table 4. In the low- and high-SUV groups, N0 patients were 15 and 2; N1-3 patients were 1 and 32; UICC stages I–II patients were 10 and 3; UICC stages III–IV patients were 6 and 31; patients with high and moderate histologic differentiation were 16 and 9; patients with poor histologic differentiation were 0 and 25. The OS and PFS of the two groups were shown in Fig. 1. Significant differences between the two groups (stages I–II and stages III–IV) were found in SUVmax (P = 0.000). The OS, PFS, and the comparison in stages I, II, III, and IV in OS were shown, respectively in Fig. 2.

Table 3 The comparison of SUVmax in patient characteristics before treatment
Full size table
Table 4 The distribution of patient characteristics in the two groups (SUVmax < 5.45 and ≥5.45)
Full size table
Fig. 1
figure 1

Comparison of the survival of two patient groups (SUVmax values <5.45 ≥5.45). a Overall survival (OS) probabilities. b Progression-free survival (PFS) probabilities

Full size image
Fig. 2
figure 2

Survival probabilities of the 50 patients with NSCLC. a Overall survival (OS) probabilities. b Overall survival (OS) probabilities of patients with UICC stages I–IV NSCLC. c Progression-free survival (PFS) probabilities

Full size image

The results of the univariate analysis were shown in Table 5: significant differences were revealed between the low- and high-SUV groups in OS and PFS (P = 0.001, 0.0005). The SUVmax, ZPS, histologic differentiation, and UICC stage were significantly associated with OS and PFS. The results of the multivariate analysis were shown in Table 6: SUVmax was significantly associated with OS (OR = 4.21, P = 0.001) and PFS (OR = 3.80, P = 0.001). Other characteristics were less associated with OS and PFS than SUVmax.

Table 5 Result of univariate analysis for the correlation with treatment outcome
Full size table
Table 6 Result of multivariate analysis for the correlation with treatment outcome
Full size table

Discussion

In our study, the SUVmax of the primary tumor was significantly correlated with tumor length, N stage, and UICC stage before treatment. And our study showed that there was a significant relationship between the SUVmax of the primary tumor and the pathologic N stage of NSCLC. Our results are consistent with those of Meng [7]. The mechanism may be interpreted as that FDG is avidly mainly taken up by tumor cells because cancer tissue consumes a large amount of glucose as an energy source [8], and the malignant cells have an altered metabolic activity with increased uptake of FDG. Tumor FDG uptake could be affected by cell differentiation, proliferative rate potential, microvessel density, and hypoxia, which are all consistent with the biological behavior of malignant cells [9, 10]. The higher the malignancy is, the greater the probability of lymph node metastasis will be, which leads to worse prognosis. A multicenter study concluded that higher SUVmax was associated with higher tumor invasion and lymph node metastasis where patients whose lung cancer showed high 18F-FDG uptake which had a 4.46- to 7.43-fold higher risk of intratumoral lymphatic vessel invasion and lymph node metastases after multivariate adjustment than those whose lung cancer showed low to moderate 18F-FDG uptake [11].

The ROC curve, used to determine the optimal cutoff value of SUVmax, was generated to assess the relationship between SUVmax and lymph node metastases (with 5.45 as the threshold), which showed a quite high sensitivity (93.9 %) and high specificity (82.4 %). In other studies, various cutoff SUV values have been used, ranging from 2.5 to 7 [12–22]. It is difficult to define an arbitrary cutoff value due to differences in patient characteristics, use of different protocols, and many other factors such as described above.

Statistical recordings of OS and PFS witnessed significant differences between the group exceeding 5.45 and the one below 5.45. When the SUVmax of the primary tumor exceeds 5.45, the OS and PFS decrease. This result is similar to our finding with regard to relationship between SUV value and N stage, UICC stage and differentiation. Analysis of Table 3 indicates that in the group with SUV values below 5.4, those patients with tumor diameter less than 5 mm accounted for 93.75 %, those at stages I and II 68.75 %, and those without lymphatic metastasis 93.75 % (including patients at stages I and II). For the group with SUV values above 5.4, those patients with maximum tumor diameter higher than 5 mm covered 97.05 % of the presence of higher T stage, N stage, and UICC stage but with mainly lower differentiated degrees.

Many studies have indicated the usefulness of FDG-PET to predict the prognosis for NSCLC patients. An European meta-analysis suggested that SUV in primary tumor before therapy had prognostic value for NSCLC [3, 23]; Higashi et al. investigated relationship between SUV and postoperative recurrence for 57 patients with resectable NSCLC through univariate analysis, reporting that patients with SUV values <5.0 outdid those with SUV values >5.0 in disease-free survival which was especially true for stage I patients (patients with SUV values <5.0 at this stage had 88 % of expected 5-year survival rate while those with SUV values >5.0 at the same stage only had 17 % of 5-year survival rate). Multivariate analyses confirmed that SUV was the most important independent prognostic factor for disease-free survival in the postoperation period [24].

Through univariate and multivariate analyses, our study found that SUVmax of the primary tumor showed statistical significance, suggesting that the FDG uptake in the primary tumor was an independent predictor for NSCLC patients (with the highest OR in SUVmax, OR = 5.13 in univariate survival analyses and OR = 4.21 in multivariate survival analyses: both are higher than those of other factors), which can be used to reflect patients’ prognosis and OS. We also found the same pattern in the PFS. The mechanism underlying this phenomenon may be explained as below: FDG-PET imaging relies on the Warburg effect, namely that cancer cells have the fundamental property of switching to anaerobic glycolysis despite of the presence of adequate oxygenation, an effect mediated in part by a hypoxia-inducible transcription factor (HIF-1) [25, 26]. HIF-1 can upregulate not only GLUT-1 and hexokinase to increase FDG uptake, but also other biological processes to promote increased tumor aggression and treatment resistance [27]. On the basis of our findings that the SUV for primary tumor predicted both local tumor control and distant metastasis, we hypothesize that tumor glucose metabolism is related to the metastatic potential of the tumor. Several investigators have also speculated that SUV is correlated with cellular proliferation or biological factors such as Ki-67, proliferating cell nuclear antigen, Glut-1, and hexokinase [28, 29].

To sum up, SUVmax may be related to tumor size, differentiation, pathological pattern, or even lymph node or distant metastasis related tumor cell proliferation and multiplication and other biological factors. Use of SUVmax for NSCLC patients have following strengths: on the one hand, SUVmax can substantially reflect malignancy degree of the tumor at molecular metastasis level by identifying individual differences of NSCLC in glycometabolism; on the other hand, SUVmax can make an objective evaluation of staging and prognosis for each NSCLC patient, provide reliable clinical basis for personalized treatment, and act as a predictive index for disease control and survival. However, it should be noted that some researches have reported results different from ours, showing that tumor size and histologic differentiation have no significant difference on multivariate analysis.

However, this prognostic value does not seem to be so clear in advanced stage lesions [30]. Hoang et al. [31] found that FDG uptake in the primary tumor of patients with advanced stage NSCLC had no prognostic value. Kieninger et al. [32] analyzed prognostic significance for early lung cancer and held that SUVmax was not an independent predictive index for OS, despite its relations with tumor staging, differentiation and other prognostic factors. Our study mainly involved stages III and IV patients (38 patients, 76 % of all patients) while stages II and III cases were in the majority, so all results and conclusions need to be further tested and verified in more experiments. Furthermore, as a semiquantitative parameter of glucose uptake, SUVmax is subject to various factors such as patient size, ROI definition, the partial-volume effect, image resolution, reconstruction methods, noise, interval time between tracer injection and imaging, attenuation correction, normalization factor, and plasma glucose level.

In conclusion, the results of our study suggest that the SUVmax from FDG-PET of primary tumors is significantly associated with OS and PFS of NSCLC. For OS and PFS, SUVmax is an independent prognostic variable. FDG uptake by the primary tumor may have prognostic value for NSCLC. For patients with higher SUV values, more aggressive adjuvant treatments should be considered. SUVmax is readily amenable to clinical use due to its non-invasive nature and convenience.