Purpose: To investigate whether 18F-fluoro-2-deoxy-glucose positron emission tomography (FDG-PET) may be a potential tool to select a subgroup of patients who might be spared maintenance treatment, if the metabolic response after first-line chemotherapy could predict time-to-progression (TTP).

Experimental Design: A total of 43 patients who underwent baseline FDG-PET scan and did not show disease progression (DP) after 4 cycles of first-line chemotherapy were enrolled and underwent second FDG-PET 3 weeks after completion of the first-line chemotherapy. The primary endpoint was to compare percent decrease in maximum standard uptake value (SUVmax) between early (TTP after second PET examination <8 weeks) and late (TTP ≥8 weeks) DP subgroups. Secondary endpoints were to determine whether fractional decrease in SUVmax could predict TTP and overall survival (OS), both calculated from the date of the second FDG-PET.

Results: Percent decreases in SUVmax in late DP subgroup were greater than those in early DP subgroup (mean reduction, 54.7% ± 27.2% vs. 27.8% ± 46.8%, P = 0.021). Receiver operating characteristic curves identified a 50.0% decrease in SUVmax as the optimal threshold to distinguish these subgroups. Using this value as the cutoff resulted in a positive predictive value of 82.6% and negative predictive value of 60.0% in predicting TTP ≥8 weeks. Patients with SUVmax decrease <50% had significantly longer median TTP (3.0 vs. 1.5 months, P = 0.001) and OS (not reached vs. 14.2 months, P = 0.003).

Conclusions: Fractional decrease in SUVmax of the main lesion after completion of 4 cycles of chemotherapy may discriminate patients with TTP ≥8 weeks and predict TTP and OS in patients with advanced NSCLC. Clin Cancer Res; 17(15); 5093–100. ©2011 AACR.

Translational Relevance

Although maintenance treatment with pemetrexed and erlotinib following first-line chemotherapy has shown positive results in patients with non–small cell lung cancer, toxicities associated with maintenance therapy are of concern in the noncurative setting. There may be a subset of patients with relatively less aggressive disease who may be overtreated by an immediate transition to maintenance treatment after first-line chemotherapy. Our findings show that metabolic response evaluated by 18F-fluoro-2-deoxy-glucose positron emission tomography (FDG-PET) using percent decrease of maximum/peak standard uptake value after completion of first-line chemotherapy may be useful to differentiate patients with longer time-to-progression after first-line chemotherapy from those with early progressive disease. In addition, good metabolic responders showed prolonged overall survival. These results suggest that FDG-PET may be a useful tool in identifying a subgroup of patients with better prognosis who may be spared immediate maintenance therapy until disease progression.

Platinum-based doublet regimens have been a mainstay of first-line treatment of patients with advanced non–small cell lung cancer (NSCLC) since a landmark meta-analysis clearly showed a significant survival benefit compared with best supportive care (1, 2). The 2009 American Society of Clinical Oncology guidelines recommended no further chemotherapy after first-line chemotherapy prior to disease progression (DP) for patients who have stable disease or who respond to first-line therapy (3, 4). The paradigm is shifting, however, following the introduction of newer, less-toxic agents including pemetrexed and erlotinib. Extending the duration of first-line chemotherapy by immediately starting early effective second-line or switch maintenance therapy with these agents has become a valid approach in the treatment of patients with advanced NSCLC (3, 4).

Concerns have arisen, however, about the toxicities and additional costs associated with chemotherapy administered in a noncurative setting to patients with a disease with poor prognosis (5). There may be a subset of patients with relatively less aggressive disease who may be overtreated by an immediate transition to further treatment after first-line therapy. To date, however, no study has reliably identified these patients (6).

Serial 18F-fluoro-2-deoxy-glucose positron emission tomography (FDG-PET) could facilitate response-adapted therapy and reduce overall healthcare costs by reducing the use of and complications associated with unnecessary treatment (7, 8). Reduction of metabolic activity after chemotherapy in advanced NSCLC was closely correlated with patient outcome (9, 10). Thus, we hypothesized that FDG-PET may be a useful determinant in identifying a subgroup of patients with better prognosis who may be spared immediate maintenance therapy until DP. To determine whether metabolic response as determined by FDG-PET, in this specific context of maintenance therapy after first-line chemotherapy, may discriminate patients with longer time-to-progression (TTP), we conducted FDG-PET before and after 4 cycles of chemotherapy in patients with stage IV NSCLC who did not progress after first-line platinum-based chemotherapy.

Patients and treatments

Forty-three patients with advanced NSCLC who underwent baseline FDG-PET and did not show evidence of DP after a fourth cycle of first-line platinum-based chemotherapy were enrolled in this study. Patients were not further treated until DP. All patients had pathologically confirmed NSCLC stage IV, as determined by the seventh edition of the American Joint Committee on Cancer staging system, measurable lesions, age 18 or older, and Eastern Cooperative Oncology Group (ECOG) performance score of 0 to 2. Enrollment of patients who received radiotherapy to nontarget lesion before primary chemotherapy was allowed. Patients having asymptomatic or treated brain metastases were eligible. Patients with previous or concurrent cancer that was distinct in primary site or histology from NSCLC, except for cervical carcinoma in situ, treated basal cell carcinoma, or superficial bladder tumors, were excluded, as were pregnant or breast-feeding women. Patients with uncontrolled diabetes were excluded. The study protocol was approved by the Institutional Review Board of the Asan Medical Center, and written informed consent was obtained from all patients.

Follow-up

FDG-PET and chest CT scans were done 3 weeks after completion of the fourth cycle of first-line chemotherapy. Following the second FDG-PET examinations, chest CT scans, along with CT scans of the abdomen and pelvis if indicated, were done every 6 weeks until progression.

FDG-PET imaging

Baseline FDG-PET was done 1 to 7 days before initiation of chemotherapy and a second FDG-PET was repeated 3 weeks after the fourth cycle of chemotherapy. Patients fasted for at least 6 hours before the examination to maintain serum glucose concentrations less than 150 mg/dL. Scanning began 60 ± 10 minutes after intravenous injection of 7.4 MBq/kg 18F-FDG. FDG-PET/CT images of skull base to the mid thigh were obtained by using PET/CT scanner [Biograph Sensation 16 (n = 19) or Biograph True Point 40 (n = 10), Siemens; or Discovery STe 8 (n = 14), GE Medical Systems]. The same scanner was used for serial scanning of the same patient. CT images for attenuation correction were acquired during shallow breathing in a spiral mode. PET images were acquired with 2 or 3 minutes of emission scan per bed for 6 to 7 bed positions with 3-dimensional acquisition mode. SUV was calculated based on injected dose and lean body weight of the patient. The volume of interest (VOI) was drawn semiautomatically over the main lung mass using vendor's software, and a maximum SUV (SUVmax) was obtained using the single maximum pixel count within the VOI. Peak SUV (SUVpeak) was defined as a mean of SUV in a sphere with a diameter of 1.5 cm centering on the SUVmax (11). Metabolic changes during chemotherapy were compared with the main lesion and used as indicators of disease status. If all lesions had disappeared, VOIs were drawn in the same areas on baseline and interim FDG-PET scans, comparing slices carefully and ensuring that the VOI was restricted to the baseline tumor.

Statistical considerations

The primary endpoint was to compare percent decrease in SUVmax in patients with TTP ≥8 weeks (late DP subgroup) and those with TTP <8 weeks (early DP subgroup). The secondary endpoints were to determine whether metabolic response assessed by percent decrease in SUVmax could predict TTP and overall survival (OS). TTP was calculated from the date of second FDG-PET examination to the date of documented progression or last follow-up. Deaths unrelated to DP were censored. OS was calculated from the date of second FDG-PET examination to the date of death from any cause or last follow-up. Both TTP and OS were calculated by using the Kaplan–Meier method. We calculated sample size by assuming that the median TTP in the early and late DP subgroups would be 6 weeks and 14 weeks, respectively. We therefore calculated that a sample size of 38 patients would provide 80% power at a 10% 2-sided significance level for patients recruited within 1 year and with a follow-up of the last patient of 14 weeks. We assumed a follow-up loss rate of 10%, thus requiring 43 patients. The Wilcoxon signed rank test was used to compare the percent decrease in SUVmax/peak between the early and late DP subgroups (12). Receiver operating characteristic (ROC) curve analysis was used to define the optimal threshold of percent SUVmax/peak decrease that could discriminate between the early and late DP subgroups (13). Kaplan–Meier curves and log-rank tests were used to assess the correlation of TTP and OS with the cutoff in percent SUVmax/peak decrease. Absolute SUVmax/peak and percent change in SUVmax/peak were expressed as means ± SDs. A value of P less than 0.05 was considered statistically significant.

Patient characteristics

Between December 2007 and November 2008, 43 patients with stage IV NSCLC were enrolled (Table 1). Except for a single case diagnosed by cytology of pleural effusion, pathologic diagnosis was confirmed by biopsy at primary lesion or metastatic lymph node. None of the patients received conventional palliative radiotherapy before primary chemotherapy. Eight patients had brain metastases and one of them underwent gamma-knife radiosurgery before chemotherapy. Patients received 4 cycles of first-line chemotherapy, consisting of gemcitabine (1,000 mg/m2 on days 1 and 8) plus cisplatin (70 mg/m2 on day 1, n = 25) or carboplatin (AUC 5 on day 1, n = 12); docetaxel (75 mg/m2 on day 1) plus cisplatin (70 mg/m2 on day 1, n = 3) or carboplatin (AUC 5 on day 1, n = 1); or paclitaxel (175 mg/m2 on day 1) plus cisplatin (70 mg/m2 on day 1, n = 2). Each cycle consisted of 3 weeks of therapy, regardless of regimen. Three patients required dose reduction of gemcitabine (n = 2) or docetaxel (n = 1) due to grade 3 neutropenia. Of these 43 patients, 28 (65.1%) showed stable disease (SD) after 4 cycles of first-line chemotherapy and 15 (34.9%) achieved a best response of partial response (PR), according to the Response Evaluation Criteria in Solid Tumors (RECIST; ref. 14).

Table 1.

Baseline patient characteristics

CharacteristicsTotalLate DP groupEarly DP groupP
(TTP ≥ 8 weeks)(TTP < 8 weeks)
Total patients 43 27 16  
Male (n, %) 31 (72.1%) 18 (66.7%) 13 (81.2%) 0.484 
Age (median, range) 59 (41–71) 56 (41–70) 61.5 (55–71) 0.073 
ECOG performance (n, %)    0.525 
 0 2 (4.7%) 2 (7.4%) 0 (0.0%)  
 1 38 (88.4%) 23 (85.2%) 15 (93.8%)  
 2 3 (7.0%) 2 (7.4%) 1 (6.2%)  
Histopathologic subtypes (n, %)    0.372 
 Squamous cell carcinoma 11 (25.6%) 5 (18.5%) 6 (37.5%)  
 Adenocarcinoma 28 (65.1%) 19 (70.4%) 9 (56.2%)  
 Unclassified 4 (9.3%) 3 (11.1%) 1 (6.2%)  
M stage (n, %)    1.000 
 M1a 8 (18.6%) 5 (18.5%) 3 (18.8%)  
 M1b 35 (81.4%) 22 (81.5) 13 (81.2%)  
Chemotherapy regimen (n, %)    0.870 
 Gemcitabine + cisplatin 25 (58.1%) 15 (55.6%) 10 (62.5%)  
 Gemcitabine + carboplatin 12 (27.9%) 8 (29.6%) 4 (25.0%)  
 Docetaxel + cisplatin 3 (7.0%) 2 (7.4%) 1 (6.2%)  
 Docetaxel + carboplatin 1 (2.3%) 1 (3.7%) 0 (0.0%)  
 Paclitaxel + cisplatin 2 (4.7%) 1 (3.7%) 1 (6.2%)  
Response to first-line chemotherapy (n, %)    0.700 
 SD 28 (65.1%) 17 (63.0%) 11 (68.8%)  
 PR 15 (34.9%) 10 (37.0%) 5 (31.2%)  
CharacteristicsTotalLate DP groupEarly DP groupP
(TTP ≥ 8 weeks)(TTP < 8 weeks)
Total patients 43 27 16  
Male (n, %) 31 (72.1%) 18 (66.7%) 13 (81.2%) 0.484 
Age (median, range) 59 (41–71) 56 (41–70) 61.5 (55–71) 0.073 
ECOG performance (n, %)    0.525 
 0 2 (4.7%) 2 (7.4%) 0 (0.0%)  
 1 38 (88.4%) 23 (85.2%) 15 (93.8%)  
 2 3 (7.0%) 2 (7.4%) 1 (6.2%)  
Histopathologic subtypes (n, %)    0.372 
 Squamous cell carcinoma 11 (25.6%) 5 (18.5%) 6 (37.5%)  
 Adenocarcinoma 28 (65.1%) 19 (70.4%) 9 (56.2%)  
 Unclassified 4 (9.3%) 3 (11.1%) 1 (6.2%)  
M stage (n, %)    1.000 
 M1a 8 (18.6%) 5 (18.5%) 3 (18.8%)  
 M1b 35 (81.4%) 22 (81.5) 13 (81.2%)  
Chemotherapy regimen (n, %)    0.870 
 Gemcitabine + cisplatin 25 (58.1%) 15 (55.6%) 10 (62.5%)  
 Gemcitabine + carboplatin 12 (27.9%) 8 (29.6%) 4 (25.0%)  
 Docetaxel + cisplatin 3 (7.0%) 2 (7.4%) 1 (6.2%)  
 Docetaxel + carboplatin 1 (2.3%) 1 (3.7%) 0 (0.0%)  
 Paclitaxel + cisplatin 2 (4.7%) 1 (3.7%) 1 (6.2%)  
Response to first-line chemotherapy (n, %)    0.700 
 SD 28 (65.1%) 17 (63.0%) 11 (68.8%)  
 PR 15 (34.9%) 10 (37.0%) 5 (31.2%)  

Patient follow-up

All 43 patients showed DP during follow-up and 41 of them were given second-line systemic treatment. Although 5 patients developed new metastases, the others showed locoregional progression with (n = 15) or without (n = 23) systemic failure. Seventeen patients died; 16 due to DP and 1 due to pneumonia. At a median follow-up period of 15.6 months (range, 9.9–22.8 months) after the second FDG-PET examination for the surviving patients, the median TTP was 2.6 months (95% CI: 2.2–2.9 months) and the median OS was 20.1 months (95% CI, 12.5–27.8 months). We found that 27 patients were in the late DP subgroup and 16 were in the early DP group. The 2 groups did not differ in patient characteristics, including gender distribution, performance status, histopathologic subtypes, M stage, and chemotherapy regimens, or in response to first-line chemotherapy (Table 1). The median TTPs in these 2 subgroups were 3.2 months (95% CI: 2.6–3.8 months) and 1.5 months (95% CI: 1.4–1.5 months), respectively (P < 0.001).

Evaluation by FDG-PET

Blood glucose levels at the first and second PET scan were 102.5 ± 14.1 mg/dL and 107.7 ± 17.7 mg/dL. Although 5 diabetic patients were included (2 in the early DP group and 3 in the late DP group), their SUVmax did not differ from those of nondiabetic patients (Supplementary Table S1A). There was no significant difference in baseline (P = 0.913) and postchemotherapy SUVmax values (P = 0.274) according to pathologic subtypes (adenocarcinoma vs. squamous cell carcinoma; Supplementary Table SA). Tumor SUVs decreased significantly after 4 cycles of first-line chemotherapy (baseline SUVmax = 11.5 ± 5.2 vs. postchemotherapy SUVmax = 6.5 ± 6.1, P < 0.001). Percent changes in SUVmax differed significantly between the early DP (27.8 ± 46.8%) and the late DP group (54.7% ± 27.2%) (P = 0.021; Table 2, Supplementary Figs. S1A and S2A). In the absence of an established cutoff value for SUVmax change, a ROC curve analysis defined a 50.0% decrease in SUVmax as the optimal threshold to distinguish the early and late DP groups. Using this value as the cutoff resulted in a sensitivity of 70.4%, a specificity of 75.0%, a positive predictive value (PPV) of 82.6%, a negative predictive value (NPV) of 60.0%, and an overall accuracy of 72.1% in predicting late DP (Table 3).

Table 2.

SUV decrease (%) in patients with TTP ≥8 weeks and <8 weeks

SUVAll patientsLate DP groupEarly DP groupP
(TTP ≥ 8 weeks)(TTP < 8 weeks)
SUVmax decrease    0.021 
 Mean ± SD 44.7% ± 37.6% 54.7% ± 27.2% 27.8 ± 46.8%  
 Median (range) 52.9% (−126.7% to 92.4%) 58.5% (−8.4% to 92.4%) 33.5% (−126.7% to 78.0%)  
Baseline SUVmax    0.741 
 Mean ± SD 11.5 ± 5.2 11.7 ± 5.9 11.1 ± 3.9  
 Median (range) 11.2 (4.2 to 28.6) 10.1 (4.4 to 28.6) 11.3 (4.2 to 19.5)  
Postchemotherapy SUVmax    0.485 
 Mean ± SD 6.5 ± 6.1 6.0 ± 7.3 7.3 ± 3.4  
 Median (range) 5.4 (1.0 to 31.0) 3.9 (1.0 to 31.0) 6.9 (2.3 to 14.4)  
SUVpeak decrease    0.039 
 Mean ± SD 40.3% ± 36.0% 48.9% ± 25.7% 25.7% ± 45.9%  
 Median (range) 41.9% (−115.9% to 88.4%) 54.8% (−4.3% to 88.4%) 33.0% (−115.9% to 75.6%)  
SUVAll patientsLate DP groupEarly DP groupP
(TTP ≥ 8 weeks)(TTP < 8 weeks)
SUVmax decrease    0.021 
 Mean ± SD 44.7% ± 37.6% 54.7% ± 27.2% 27.8 ± 46.8%  
 Median (range) 52.9% (−126.7% to 92.4%) 58.5% (−8.4% to 92.4%) 33.5% (−126.7% to 78.0%)  
Baseline SUVmax    0.741 
 Mean ± SD 11.5 ± 5.2 11.7 ± 5.9 11.1 ± 3.9  
 Median (range) 11.2 (4.2 to 28.6) 10.1 (4.4 to 28.6) 11.3 (4.2 to 19.5)  
Postchemotherapy SUVmax    0.485 
 Mean ± SD 6.5 ± 6.1 6.0 ± 7.3 7.3 ± 3.4  
 Median (range) 5.4 (1.0 to 31.0) 3.9 (1.0 to 31.0) 6.9 (2.3 to 14.4)  
SUVpeak decrease    0.039 
 Mean ± SD 40.3% ± 36.0% 48.9% ± 25.7% 25.7% ± 45.9%  
 Median (range) 41.9% (−115.9% to 88.4%) 54.8% (−4.3% to 88.4%) 33.0% (−115.9% to 75.6%)  
Table 3.

SUV decrease of 50% as a cutoff value for prediction of TTP ≥8 weeks

SUV decreaseLate DPEarly DPTotalPSensitivitySpecificityPPVNPVAccuracy
(n = 27)(n = 16)
SUVmax decrease ≥50% 19 23 0.005 70.4% 75.0% 82.6% 60.0% 72.1% 
SUVmax decrease <50% 12 20       
SUVpeak decrease ≥50% 16 20 0.029 59.3% 75.0% 80.0% 52.2% 65.1% 
SUVpeak decrease <50% 11 12 23       
SUV decreaseLate DPEarly DPTotalPSensitivitySpecificityPPVNPVAccuracy
(n = 27)(n = 16)
SUVmax decrease ≥50% 19 23 0.005 70.4% 75.0% 82.6% 60.0% 72.1% 
SUVmax decrease <50% 12 20       
SUVpeak decrease ≥50% 16 20 0.029 59.3% 75.0% 80.0% 52.2% 65.1% 
SUVpeak decrease <50% 11 12 23       

TTP and OS according to FDG-PET metabolic responses

When we sorted patients based on a 50% decrease in SUVmax, we found that 23 patients were good metabolic responders (≥50.0% decrease in SUVmax) and 20 were poor metabolic responders (<50% decrease in SUVmax). The 2 groups did not significantly differ in patient characteristics (Supplementary Table S3A). The median TTPs in these 2 subgroups were 3.0 months (95% CI: 2.3–3.7 months) and 1.5 months (95% CI: 1.4–1.7 months), respectively (HR, 0.27; 95% CI: 0.13–0.57, P = 0.001; Fig. 1A). While a median OS was not reached in the good metabolic responders, it was 14.2 months (95% CI: 6.0–22.4 months) in the poor metabolic responders (HR 0.21, 95% CI: 0.08–0.58, P = 0.003; Fig. 1B)

Figure 1.

TTP (A) and OS (B) according to fractional decrease in SUVmax (<50% vs. ≥50%).

Figure 1.

TTP (A) and OS (B) according to fractional decrease in SUVmax (<50% vs. ≥50%).

Close modal

Prognostic factors

Female gender (vs. male) and stage M1a (vs. M1b) were significantly associated with a longer OS (P = 0.012 and P = 0.037, respectively) on univariate analyses. No other baseline demographic or clinical characteristics were significantly associated with TTP or OS in the univariate analysis (Table 4). When a multivariate analysis was done with these significant predictors of OS in univariate analyses, a 50% reduction in SUVmax was the only significant predictor of TTP (HR, 0.317; 95% CI: 0.158–0.639; P = 0.001) and OS (HR, 0.251; 95% CI: 0.079–0.799; P = 0.019; Table 5, Supplementary Table S4A).

Table 4.

Univariate analysis for TTP and OS

Prognostic factorsnMedian TTP (95% CI)PMedian OS (95% CI)P
Age, y   0.810  0.073 
 <65 30 2.566 (1.474–3.658)  20.132 (ND)  
 ≥65 13 2.895 (1.813–3.976)  13.322 (9.614–17.031)  
Sex   0.873  0.012 
 Male 31 2.566 (1.310–3.922)  14.211 (9.707–18.714)  
 Female 12 2.566 (2.007–3.124)  20.132 (ND)  
Pathologic subtype   0.191  0.241 
 Adenocarcinoma 28 2.829 (2.319–3.339)  NR  
 Nonadenocarcinoma 15 2.500 (1.317–3.683)  14.211 (3.443–24.978)  
M stage   0.960  0.037 
 M1a 2.829 (0.960–4.698)  NR  
 M1b 22 2.566 (1.964–3.167)  17.763 (13.229–22.297)  
First-line regimen   0.340  0.720 
 Taxane based 2.533 (1.388–3.678)  20.132 (7.708–32.555)  
 Gemcitabine based 37 2.697 (1.946–3.449)  17.763 (ND)  
Response to first-line chemotherapy   0.201  0.478 
 PR 15 2.500 (1.296–3.704)  20.132 (16.484–23.779)  
 SD 28 2.566 (2.259–2.873)  NR  
Decrease in SUVmax   0.001  0.003 
 ≥50% 23 3.026 (2.306–3.747)  NR  
 <50% 20 1.546 (1.366–1.726)  14.178 (5.967–22.388)  
Decrease in SUVpeak   0.012  0.012 
 ≥50% 20 2.961 (2.672–3.249)  NR  
 <50% 23 1.678 (0.185–3.171)  14.178 (8.006–20.349)  
Prognostic factorsnMedian TTP (95% CI)PMedian OS (95% CI)P
Age, y   0.810  0.073 
 <65 30 2.566 (1.474–3.658)  20.132 (ND)  
 ≥65 13 2.895 (1.813–3.976)  13.322 (9.614–17.031)  
Sex   0.873  0.012 
 Male 31 2.566 (1.310–3.922)  14.211 (9.707–18.714)  
 Female 12 2.566 (2.007–3.124)  20.132 (ND)  
Pathologic subtype   0.191  0.241 
 Adenocarcinoma 28 2.829 (2.319–3.339)  NR  
 Nonadenocarcinoma 15 2.500 (1.317–3.683)  14.211 (3.443–24.978)  
M stage   0.960  0.037 
 M1a 2.829 (0.960–4.698)  NR  
 M1b 22 2.566 (1.964–3.167)  17.763 (13.229–22.297)  
First-line regimen   0.340  0.720 
 Taxane based 2.533 (1.388–3.678)  20.132 (7.708–32.555)  
 Gemcitabine based 37 2.697 (1.946–3.449)  17.763 (ND)  
Response to first-line chemotherapy   0.201  0.478 
 PR 15 2.500 (1.296–3.704)  20.132 (16.484–23.779)  
 SD 28 2.566 (2.259–2.873)  NR  
Decrease in SUVmax   0.001  0.003 
 ≥50% 23 3.026 (2.306–3.747)  NR  
 <50% 20 1.546 (1.366–1.726)  14.178 (5.967–22.388)  
Decrease in SUVpeak   0.012  0.012 
 ≥50% 20 2.961 (2.672–3.249)  NR  
 <50% 23 1.678 (0.185–3.171)  14.178 (8.006–20.349)  

Abbreviations: ND, not defined; NR, not reached.

Table 5.

Multivariate analysis for TTP and OS

Prognostic factorsTTPOS
PHR95% CIPHR95% CI
Sex (female vs. male) 0.767 1.121 (0.527–2.385) 0.196 0.256 (0.033–2.021) 
M stage (M1a vs. M1b) 0.880 0.936 (0.394–2.224) 0.973 NDa NDa 
Decrease in SUVmax (≥ 50% vs. < 50%) 0.001 0.317 (0.158–0.639) 0.019 0.251 (0.079–0.799) 
Prognostic factorsTTPOS
PHR95% CIPHR95% CI
Sex (female vs. male) 0.767 1.121 (0.527–2.385) 0.196 0.256 (0.033–2.021) 
M stage (M1a vs. M1b) 0.880 0.936 (0.394–2.224) 0.973 NDa NDa 
Decrease in SUVmax (≥ 50% vs. < 50%) 0.001 0.317 (0.158–0.639) 0.019 0.251 (0.079–0.799) 

Abbreviation: ND, not defined.

aNone of the 8 patients with M1a died during the follow-up period.

Exploratory analysis with SUVpeak

We underwent exploratory analysis with SUVpeak. Fractional decrease in SUVpeak after primary chemotherapy was significantly higher in the late DP group (48.9% ± 25.7% vs. 25.7% ± 45.9%, P = 0.039; Table 2, Supplementary Fig. S1A). Decrease in SUVpeak of 50.0% was the optimal cutoff to differentiate the early and late DP groups in a ROC curve analysis resulting in PPV of 80.0% predicting late DP (Table 3). Twenty patients showed superior metabolic response (≥50.0% decreases in SUVpeak). Decrease in SUVpeak by 50% was a significant predictor of both TTP (P = 0.012) and OS (P = 0.012; Table 4 and Supplementary Fig. S3A). When a multivariate analysis was done with gender (male vs. female) and M stage (M1a vs. M1b), a 50% reduction in SUVpeak was significantly associated with TTP (HR, 0.403; 95% CI: 0.200–0.816; P = 0.012) and a predictor of OS with marginal statistical significance (HR, 0.347; 95% CI: 0.112–1.074; P = 0.066; Supplementary Table S5A).

A second FDG-PET examination after first-line chemotherapy seems useful in discriminating early and late DP subgroups and predicting both TTP and OS in patients with advanced NSCLC. Fractional decrease in SUVmax 50% or more after first-line chemotherapy could distinguish a late DP subgroup with a PPV of 82.6%. Furthermore, those patients who showed a 50% or more decrease in SUVmax of the main lesion progressed later (median TTP, 3.0 vs. 1.6 months after second FDG-PET, P = 0.001) and survived longer (median OS, not reached vs. 14.2 months after second FDG-PET, P = 0.003) than those with a less 50% decrease in SUVmax.

Maintenance treatment with pemetrexed or erlotinib following first-line treatment has improved OS in patients with NSCLC (3, 4, 15). Although this strategy seems appealing, toxicities associated with chemotherapy are of concern in noncurative settings (5). Despite favorable toxicity profiles of the agents, immediate maintenance chemotherapy with the drugs was associated with higher incidences of drug-related grade 3/4 toxicities (16% vs. 4% in the pemetrexed maintenance trial, P < 0.0001; 12% vs. 1% in SATURN trial involving maintenance erlotinib; ref. 3, 4). Furthermore, the trials included a controversial placebo arm; one-third of these patients did not receive second-line treatment and only some of them were given the maintenance agents: 18% in the pemetrexed trial and 21% of either erlotinib or gefitinib in SATURN trial, respectively. The OS for patients in the delayed docetaxel arm who actually received docetaxel was similar to that in the immediate docetaxel arm in Fidias' study (16). These findings raise the question of whether the survival benefit would have been preserved if more patients in the placebo arm had received second-line treatments. It is therefore not clear whether optimum treatment should involve use of this maintenance strategy or whether it is better to wait until progression of disease to start second-line therapy. However, the Food and Drug Administration approved both pemetrexed and erlotinib for the purpose and this therapeutic approach is also listed in current National Comprehensive Cancer Network guidelines for the treatment of lung cancer (v3.2011). The strategy of immediate maintenance therapy is presently adopted by at least a party of physicians. It would be therefore worthwhile to identify patients who do and do not require immediate maintenance therapy.

A meta-analysis showed that the metabolic activity of the primary tumor, as determined by SUV on FDG-PET imaging, was a prognostic factor in patients with NSCLC (17). Fractional decreases in SUV following chemotherapy, ranging from 20% to 50%, were also able to prognostically stratify these patients (9, 10, 18, 19). One issue that must be addressed is the optimal timing of serial FDG-PET. In patients with locally advanced NSCLC, interim FDG-PET after 1 cycle of neoadjuvant chemotherapy was able to discriminate among patients with different prognoses (9, 20). However, FDG-PET after 3 cycles of chemotherapy showed the highest correlation with the final outcome of therapy (20, 21). Because FDG uptake by a tumor changes according to treatment response, we hypothesized that a second FDG-PET examination after the completion of 4 cycles of first-line chemotherapy would be the most reliable measurement of metabolic tumor response in predicting TTP. However, none of the studies, to our knowledge, have investigated the role of FDG-PET in selecting patients for maintenance therapy and our study showed that FDG-PET scan results could identify patients with TTP ≥8 weeks and <8 weeks. We chose an 8-week cutoff because 6 to 8 weeks is the usual timing for evaluation after first-line chemotherapy and a difference in TTP between the maintenance and observation arms ranged from 1 to 3 months (3, 4, 15), making it the optimal minimum chemotherapy holiday.

A 50% or more decrease in SUVmax, as the threshold for a FDG-PET response, could differentiate patients into early and late DP subgroups, with a PPV of 82.6% for the prediction of late DP. This indicates that at least more than 80% of patients do not progress early without maintenance chemotherapy if their fractional decrease in SUVmax exceeds 50%. Close follow-up of patients, however, is still required not to lose the opportunity of second-line treatment in those who progress early despite good metabolic responses. We also found that both median TTP and OS were significantly longer in the FDG-PET responders than in nonresponders, while anatomic response according to RECIST (i.e., SD vs. PR) could not predict both TTP and OS. In addition to SUVmax, we conducted exploratory analysis with SUVpeak, which is the average SUV within a finite volume around the maximum pixel value as the concept was introduced in the PET Response Criteria in Solid Tumors in 2009 (11). Fractional decrease in SUVpeak was also significantly higher in the late DP group and 50% decrease in SUVpeak following primary chemotherapy could distinguish the late DP from the early DP group with a PPV of 80.0%, which are consistent with results from SUVmax change. Taken together, evaluation of metabolic response by FDG-PET may be a potential tool in identifying patients with longer TTP who may be overtreated by an immediate transition to further treatment after first-line therapy. Thus, a 50% or more decrease in SUVmax or SUVpeak may be relevant reference criteria for future studies. This fractional change in SUVmax or SUVpeak may be a better cutoff than absolute value because the SUV thresholds identified in single-institution studies involving a specific set of patients may not be applicable to other institutions with different equipment, patient populations, and clinical imaging protocols (22).

Our findings suggest that comparison of FDG-PET before and after first-line chemotherapy can discriminate a subgroup of patients with longer TTP and predict both TTP and OS in patients with advanced NSCLC using a specific cutoff value of percent change in SUVmax or SUVpeak. This approach may represent a novel application of FDG-PET which has not been studied before. We defined the metabolic response criteria for prospective trials. However, validation studies are required because we used a post hoc definition of the criteria for predicting TTP. Future randomized trials of maintenance therapy based on metabolic responses, as assessed by FDG-PET, may better define the role of FDG-PET in selecting a patient subpopulation that may be spared immediate maintenance chemotherapy.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Non-small Cell Lung Cancer Collaborative Group
. 
Chemotherapy in non-small cell lung cancer: a meta-analysis using updated data on individual patients from 52 randomised clinical trials
.
BMJ
1995
;
311
:
899
909
.
2.
Socinski
MA
. 
Re-evaluating duration of therapy in advanced non-small-cell lung cancer: is it really duration or is it more about timing and exposure?
J Clin Oncol
2009
;
27
:
3268
70
.
3.
Ciuleanu
T
,
Brodowicz
T
,
Zielinski
C
,
Kim
JH
,
Krzakowski
M
,
Laack
E
, et al
Maintenance pemetrexed plus best supportive care versus placebo plus best supportive care for non-small-cell lung cancer: a randomised, double-blind, phase 3 study
.
Lancet
2009
;
374
:
1432
40
.
4.
Cappuzzo
F
,
Ciuleanu
T
,
Stelmakh
L
,
Cicenas
S
,
Szczésna
A
,
Juhász
E
, et al
Erlotinib as maintenance treatment in advanced non-small-cell lung cancer: a multicentre, randomised, placebo-controlled phase 3 study
. Lancet Oncol 
2010
;
11
:
521
9
.
5.
Schiller
JH
,
Ramalingam
SS
. 
Duration of chemotherapy for metastatic non-small-cell lung cancer: more may be better after all
.
J Clin Oncol
2009
;
27
:
3265
7
.
6.
Stinchcombe
TE
,
West
HL
. 
Maintenance therapy in non-small-cell lung cancer
.
Lancet
2009
;
374
:
1398
400
.
7.
Nahmias
C
,
Hanna
WT
,
Wahl
LM
,
Long
MJ
,
Hubner
KF
,
Townsend
DW
. 
Time course of early response to chemotherapy in non-small cell lung cancer patients with 18F-FDG PET/CT
.
J Nucl Med
2007
;
48
:
744
51
.
8.
Hicks
RJ
. 
Role of 18F-FDG PET in assessment of response in non-small cell lung cancer
.
J Nucl Med
2009
;
50
:
31S
42
.
9.
Weber
WA
,
Petersen
V
,
Schmidt
B
,
Tyndale-Hines
L
,
Link
T
,
Peschel
C
, et al
Positron emission tomography in non-small-cell lung cancer: prediction of response to chemotherapy by quantitative assessment of glucose use
.
J Clin Oncol
2003
;
21
:
2651
7
.
10.
de Geus-Oei
LF
,
van der Heijden
HF
,
Visser
EP
,
Hermsen
R
,
van Hoorn
BA
,
Timmer-Bonte
JN
, et al
Chemotherapy response evaluation with 18F-FDG PET in patients with non-small cell lung cancer
.
J Nucl Med
2007
;
48
:
1592
8
.
11.
Wahl
RL
,
Jacene
H
,
Kasamon
Y
,
Lodge
MA
. 
From RECIST to PERCIST: evolving considerations for PET response criteria in solid tumors
.
J Nucl Med
2009
;
50
Suppl 1
:
122S
50
.
12.
Wilcoxon
F
. 
Individual comparisons by ranking methods
.
Biometrics Bulletin
1945
;
1
:
80
3
.
13.
Metz
CE
. 
Some practical issues of experimental design and data analysis in radiological ROC studies
.
Invest Radiol
1989
;
24
:
234
45
.
14.
Therasse
P
,
Arbuck
SG
,
Eisenhauer
EA
,
Wanders
J
,
Kaplan
RS
,
Rubinstein
L
, et al
New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada
.
J Natl Cancer Inst
2000
;
92
:
205
16
.
15.
Gridelli
C
,
Ardizzoni
A
,
Douillard
JY
,
Hanna
N
,
Manegold
C
,
Perrone
F
, et al
Recent issues in first-line treatment of advanced non-small-cell lung cancer: Results of an International Expert Panel Meeting of the Italian Association of Thoracic Oncology
.
Lung Cancer
2010
;
68
:
319
31
.
16.
Fidias
PM
,
Dakhil
SR
,
Lyss
AP
,
Loesch
DM
,
Waterhouse
DM
,
Bromund
JL
, et al
Phase III study of immediate compared with delayed docetaxel after front-line therapy with gemcitabine plus carboplatin in advanced non-small-cell lung cancer
.
J Clin Oncol
2009
;
27
:
591
8
.
17.
Berghmans
T
,
Dusart
M
,
Paesmans
M
,
Hossein-Foucher
C
,
Buvat
I
,
Castaigne
C
, et al
Primary tumor standardized uptake value (SUVmax) measured on fluorodeoxyglucose positron emission tomography (FDG-PET) is of prognostic value for survival in non-small cell lung cancer (NSCLC): a systematic review and meta-analysis (MA) by the European Lung Cancer Working Party for the IASLC Lung Cancer Staging Project
.
J Thorac Oncol
2008
;
3
:
6
12
.
18.
Pöttgen
C
,
Levegrün
S
,
Theegarten
D
,
Marnitz
S
,
Grehl
S
,
Pink
R
, et al
Value of 18F-fluoro-2-deoxy-D-glucose-positron emission tomography/computed tomography in non-small-cell lung cancer for prediction of pathologic response and times to relapse after neoadjuvant chemoradiotherapy
.
Clin Cancer Res
2006
;
12
:
97
106
.
19.
Vansteenkiste
JF
,
Stroobants
SG
,
De Leyn
PR
,
Dupont
PJ
,
Verbeken
EK
The Leuven Lung Cancer Group
. 
Potential use of FDG-PET scan after induction chemotherapy in surgically staged IIIa-N2 non-small-cell lung cancer: a prospective pilot study
.
Ann Oncol
1998
;
9
:
1193
8
.
20.
Hoekstra
CJ
,
Stroobants
SG
,
Smit
EF
,
Vansteenkiste
J
,
van Tinteren
H
,
Postmus
PE
, et al
Prognostic relevance of response evaluation using [18F]-2-fluoro-2-deoxy-D-glucose positron emission tomography in patients with locally advanced non-small-cell lung cancer
.
J Clin Oncol
2005
;
23
:
8362
70
.
21.
Shankar
LK
,
Hoffman
JM
,
Bacharach
S
,
Graham
MM
,
Karp
J
,
Lammertsma
AA
, et al
Consensus recommendations for the use of 18F-FDG PET as an indicator of therapeutic response in patients in National Cancer Institute Trials
.
J Nucl Med
2006
;
47
:
1059
66
.
22.
Schoder
H
,
Fury
M
,
Lee
N
,
Kraus
D
. 
PET monitoring of therapy response in head and neck squamous cell carcinoma
.
J Nucl Med
2009
;
50
Suppl 1
:
74S
88
.