Prognostic Relevance of Response Evaluation Using [18F]-2-Fluoro-2-Deoxy-D-Glucose Positron Emission Tomography in Patients With Locally Adv
http://www.100md.com
《临床肿瘤学》
the Departments of Nuclear Medicine and Positron Emission Tomography Research, Pulmonary Medicine, and Radiology, Vrije University Medical Center, Amsterdam
Department of Pulmonary Medicine, Jeroen Bosch Hospital, 's-Hertogenbosch
Department of Pulmonary Medicine, St Antonius Hospital, Nieuwegein
Department of Thoracic Oncology, Cancer Institute/Antoni van Leeuwenhoek Hospital
Comprehensive Cancer Center, Amsterdam, the Netherlands
Departments of Nuclear Medicine and Pulmonary Medicine, University Hospital Gasthuisberg, Leuven, Belgium
ABSTRACT
PURPOSE: The objective of this study was to determine the accuracy of (early) response measurements using [18F]-2-fluoro-2-deoxy-D-glucose positron emission tomography (18FDG PET) with respect to survival of patients with stage IIIA-N2 non–small-cell lung cancer (NSCLC) undergoing induction chemotherapy (IC), with a comparative analysis of PET methods.
PATIENTS AND METHODS: In a prospective multicenter study, PET was performed in patients before IC and after one and three cycles. Computed tomography (CT) was performed before and after IC. Glucose consumption (metabolic rate of glucose [MRglu]) was measured using Patlak graphical analysis and correlated with simplified methods. Mediastinal lymph node (MLN) status was assessed visually. Cox proportional hazards analysis was used to determine the prognostic relevance of CT and PET measures of response with respect to survival.
RESULTS: Complete PET data sets were available in 47 patients. Median survival was 21 months. MLN status after IC by PET predicted survival (hazard ratio [HR], 2.33; 95% CI, 1.04 to 5.22; P = .04) in contrast with CT (HR, 1.87; 95% CI, 0.81 to 4.30; P = .14). Residual MRglu after IC proved to be the best prognostic factor (HR, 1.95; 95% CI, 1.28 to 2.97; P = .002). Multivariate stepwise analysis showed that PET identified prognostically different strata in patients considered responsive according to CT. Residual MRglu after one cycle selected patients with different outcomes (HR, 2.04; 95% CI, 1.18 to 3.52; P = .01). Simplified quantitative 18FDG PET methods were correlated with Patlak graphical analysis during and after therapy (r 0.90).
CONCLUSION: 18FDG PET has additional value over CT in monitoring response to IC in patients with stage IIIA-N2 NSCLC, and it seems feasible to predict survival early during IC. Simple semiquantitative and complex PET methods perform equally well.
INTRODUCTION
At present, monitoring response to induction chemotherapy (IC) in non–small-cell lung cancer (NSCLC) is a post hoc approach; in other words, at the end of treatment, volumetric changes of known tumor sites are measured.1 In locally advanced NSCLC (stage IIIA-N2), the clinical utility of standard computed tomography (CT) criteria to document response to IC is limited.2,3 Although retrospective studies have suggested that radiologic response to IC carries prognostic information with respect to survival,4 in the majority of patients, information with indeterminate prognostic relevance is obtained, as typical response rates (mostly partial) to IC are approximately 70%.5 A more reliable stratification of patients subjected to combined-modality therapy (CMT) could have major clinical implications, especially if prognostically relevant information would be available at an early stage of induction treatment.
Positron emission tomography (PET) assesses the effect of therapy at a metabolic rather than at an anatomic level. [18F]-2-fluoro-2-deoxy-D-glucose (18FDG) has high affinity for NSCLC, and as such, whole-body PET improves the selection of patients eligible for CMT.6-8
There are some data to suggest that changes of glucose metabolism as measured with 18FDG PET after completed therapy reflect therapeutic effectiveness more accurately than volumetric measures.6,9-13 However, assessment of the value of 18FDG PET for monitoring response is complicated by heterogeneity of published data with respect to the applied methods of PET quantification (comprising an array of visual and [semi-] quantitative measures14), the primary targets of PET evaluation (primary tumor, lymph node status), and the clinical end points (histology, survival). As a result, objective validation of proposed thresholds of responsiveness for PET is lacking. It is unclear whether and how application of 18FDG PET improves the present standard of radiologic response classification. Finally, the potential of therapy evaluation with PET at an earlier and perhaps clinically more relevant stage of NSCLC treatment has only been demonstrated for stage IV NSCLC.6
The objective of the present study was to investigate the accuracy of (early) response evaluation using 18FDG PET to predict survival of patients undergoing CMT for stage IIIA-N2 NSCLC using a study design, which allowed for comparative analysis of quantitative PET methods at various levels of complexity.
PATIENTS AND METHODS
Between January 1998 and May 2001, this multicenter study was open for accrual. Before PET, patients had been staged as pIIIA-N2 NSCLC and were considered candidates for CMT. Initial staging was performed according to the guidelines of the American Society of Clinical Oncology.15 Patients with insulin-dependent diabetes mellitus were excluded. PET scans were performed at the clinical PET center of the Vrije University Medical Center, Amsterdam, the Netherlands, or the University Hospital Gasthuisberg, Leuven, Belgium. All patients received three cycles of platinum-based IC, usually containing a third-generation cytotoxic agent (Fig 1). In the Netherlands, patients without progression of disease were subsequently randomly assigned to either surgery or radiotherapy according to study protocol 08941 by the European Organization for Research and Treatment of Cancer (EORTC; available at www.eortc.be). In Belgium, patients were treated in a prospective protocol of vindesine, ifosfamide, and platinum induction followed by attempted complete resection,16 whereas radical radiotherapy was applied only in case of unacceptable surgical risk because of cardiopulmonary limitations, when the disease was considered irresectable postinduction by the attending thoracic surgeon, or on patient's request. The medical ethics committees of all participating institutes approved the study protocol. All patients gave informed consent before inclusion according to local medical ethical committee regulations.
PET scans were performed after conventional staging was completed. Dynamic PET scans were performed before (1 to 2 weeks) the start of IC, before the second cycle (1 to 5 days), and within 3 to 4 weeks after completion of the third cycle of IC, but before the start of locoregional therapy. In addition, we performed whole-body scans before IC and after the third cycle. Clinicians were blinded to the results of the dynamic PET scans but were informed about unexpected metastatic findings on the whole-body scans, in the best interest of the patient. In the context of patient care, upstaging by PET had to be confirmed with conventional imaging or pathology before altering therapy.17
Radiotherapy
The aim of radiotherapy was to irradiate the primary tumor and areas of radiologic involvement with at least 2-cm margins to a dose of 60.0 to 62.5 Gy, using daily fractions of 2 Gy, once per day, 5 days a week. The lymphatic drainage areas of the mediastinum, which were not abnormal on CT, received a prophylactic dose of 40 to 45 Gy in daily fractions of 2 Gy.
Surgery
Thoracotomy with systematic nodal dissection was performed with the aim of complete resection of the primary tumor, ipsilateral hilar nodes, and mediastinal lymph nodes. Even in case of extensive nodal involvement, complete resection was attempted. Complete resection was defined as macroscopic and microscopic free tumor margins. The highest mediastinal node had to be free of tumor. In case of unsuspected metastasis, or when it was clear that resection would be macroscopically (and proven by frozen section) incomplete, only systematic mediastinal lymphadenectomy for staging purposes was performed, provided that the per-operative condition of the patient allowed this.
PET Acquisition
Scans were performed using state-of-the-art PET scanners (ECAT EXACT HR+; Siemens/CTI, Knoxville, TN). These scanners have an axial field of view of 15 cm, divided into 63 contiguous planes. Patients were positioned supine on the scanner bed with the tumor in the center of the axial field of view of the scanner. Patients had fasted for at least 6 hours before scanning.
Acquisition started with a 10- to 15-minute transmission scan to correct for photon attenuation,18 followed by a bolus injection of 370 MBq of 18FDG in 5 mL of saline through an injector (Medrad International, Maastricht, the Netherlands) at a rate of 0.8 mL/sec, after which the line was flushed with 42 mL of saline (2.0 mL/sec). Simultaneously with the injection of 18FDG, a dynamic emission scan (in two-dimensional mode) was started with a total duration of 60 minutes with variable frame length (6 x 5, 6 x 10, 3 x 20, 5 x 30, 5 x 60, 8 x 150, and 6 x 300 seconds). All dynamic scan data were corrected for dead time, decay, scatter, randoms, and photon attenuation and were reconstructed as 128 x 128 matrices using filtered back projection (FBP) with a Hanning filter (cutoff, 0.5 cycles/pixel). This resulted in a transaxial spatial resolution of around 7-mm full width at half maximum. In addition, three venous blood samples were drawn at 35, 45, and 55 minutes after 18FDG injection as quality control for the image-derived input function19,20 and for plasma glucose measurement (hexokinase method, Hitachi 747; Boeringer Mannheim, Mannheim, Germany).
Data Analysis
An experienced radiologist blinded to patient data assessed tumor response. Response was based on the comparison of two-dimensional measurable lesions on a chest spiral CT scan obtained before and after three cycles of IC using WHO criteria.15 Persistent N2 disease after IC was defined on CT as persistence of enlarged MLN (> 1 cm in the short axis).
Quantitative PET data analysis was performed without any knowledge of clinical outcome. Using 18FDG PET data, three-dimensional regions of interest (ROIs) were defined semiautomatically over the tumor using a threshold of 50% of the maximum pixel value within the tumor after background correction (using a manually drawn ROI over normal tissue next to the tumor).21,32 The (local) background correction is applied to obtain realistic volumes of interest (VOI) in case of low contrast between tumor and background, and should, in theory, most closely correspond to actual tumor volume. In fact, it provides a correction for the limited resolution of PET images. In short, first the isocontour at 70% of tumor maximum is defined automatically as a first guess of tumor extent. Next, a three-dimensional shell of 1 voxel thickness at approximately 20 mm (three times the image resolution) distance from the outer voxels of the 70% VOI was defined. The average voxel value of the voxels within this shell is used as an estimate of average local background. Finally, background corrected VOIs are generated by taking the 50% value of tumor maximum minus this background value. For this purpose the last three frames of the sinograms (45 of 60 minutes postinjection) were summed and reconstructed using ordered subset expectation maximization with two iterations and 12 subsets followed by postsmoothing of the reconstructed image using a 5-mm full width at half maximum Gaussian filter was used to obtain the same resolution as FBP data.22 In addition, using FBP data, ROIs were defined manually over the aortic arch, left ventricle, and left atrium to obtain an image-derived input function, as described previously.19 MLN metastases were analyzed visually; focally enhanced activity above mediastinal background was considered positive.
Initially, response analysis of the primary tumor was performed using nonlinear regression (NLR), using the standard two-tissue compartment model with three (3k) rate constants, a blood volume component, and an image-derived input function. A known disadvantage of NLR is its sensitivity to noise, which will increase for low values of glucose consumption (MRglu) in a lesion (ie, after good response). In a previous study,23 a high correlation between NLR and Patlak graphical analysis at 10 to 60 minutes postinjection (a method less susceptible to noise in the data) was found. A correlation analysis for the present data showed improvement in this correlation (r = 0.98 to 0.99) when leaving out a few obvious outliers in the NLR data. On the basis of this observation, Patlak graphical analysis was used to analyze the study data. In addition, two simplified methods were used23: the standard uptake value (SUV) at 40 to 60 minutes postinjection (corrected for body-surface area and plasma glucose; SUVBSAg), and the simplified kinetic method (SKM) as described by Hunter et al24 at 40 to 60 minutes postinjection. All methods were applied to the same data set. A more detailed description of these methods can be found elsewhere.14 The lumped constant, which accounts for differences between glucose and 18FDG, was set to 1 and assumed to be constant over time, as no studies on the actual value of the lumped constant in tumors outside the CNS have been reported.14
Follow-Up
Every 2 months after definitive locoregional therapy, patients underwent blood tests, physical examination, and chest radiography. When tumor relapse was suspected, additional tests were performed.
Statistics
The correlation of percentage change in MRglu over time with Patlak was assessed. Cox proportional hazards regression analysis was used to assess the prognostic value of the baseline and absolute uptake after one and three courses, respectively, and the change relative to baseline MRglu and after one and three courses with respect to survival time.
Results are presented by hazard ratios with 95% CIs. Basic assumptions of linearity and additivity were checked. The value of MLN status as assessed by CT and 18FDG PET were analyzed using Kaplan and Meier plots and the log-rank test (P < .05). Finally, models were constructed of variables related to baseline and change in uptake after one cycle, and these were compared by means of the Akaike Information Criterion.25 Cut points were established by means of exploring martingale residual plots and the minimum P value approach.
MRglu results at baseline and after one and three courses of IC by Patlak analysis were then correlated with SUVBSAg and SKM results.
RESULTS
Seventy-nine patients were included in the study. After the first PET scan, but before IC, 23 patients were excluded because of confirmed upstaging (stage IIIB/IV, n = 12), technical problems (n = 9), or patient refusal (n = 2). After the second PET scan (after one IC cycle), another nine patients were excluded because of technical problems (n = 4), exclusion from the EORTC protocol (n = 4), or refusal to undergo a third PET scan (n = 1; Fig 2).
Therefore, complete PET data sets were available in 47 patients (Table 1) to be used for analysis of prognostic value. After induction chemotherapy, 25 patients were treated with surgery, 15 patients were treated with radical radiotherapy, and six patients were treated with radiotherapy with palliative intent because of progression based on CT criteria during IC. In the remaining patient, no locoregional therapy was given, as response to IC was unclear and his physical condition deteriorated. Resection was radical in 21 patients. Irradical resection as a result of persisting MLN involvement was found in three patients, and in one patient, the primary tumor could not be radically removed.
Median follow-up of all patients was 28 months, with a minimum of 10 months in surviving patients. During the first 2 years of follow-up, 24 patients died of disease, and recurrent disease was established in five other patients. Median survival was 21 months. In 10 patients, PET had suggested N3 (n = 2) or distant (n = 8) metastatic disease, which could not be confirmed clinically. According to the study protocol, these patients were treated with multimodality therapy, and their overall survival was similar to that of the remaining patients (P = .74).
Prognostic Indicators: After Induction Therapy
At the end of IC, N stage at CT was not predictive of survival (log-rank test, P = .14), but focally enhanced MLN 18FDG uptake was associated with a two-fold higher risk of mortality (log-rank test, P = .04; Fig 3). In the operated subset (n = 25), sensitivity and specificity of 18FDG PET after IC for MLN (N2) involvement were 50% (95% CI, 19% to 81%) and 71% (95% CI, 42% to 92%), respectively (corresponding to positive and negative predictive values of 66% [95% CI, 21% to 86%] and 67% [95% CI, 38% to 88%], respectively).
Using standard WHO criteria, the majority of patients (n = 26) had a partial response after completed IC, six patients had a minor response, seven patients had stable disease, five patients had progressive disease, and two patients showed complete response at CT. In one patient, response could not be classified as no CT scan was performed before the start of IC.
The median fractional change of metabolic rate of glucose (MRglu) with respect to baseline was –48% (range, –28 to –97%; Fig 4). The previously defined PET criterion for response (50% decrease of [18F]FDG uptake after three cycles v baseline11) either expressed as MRglu or SUVBSAg predicted survival significantly (log-rank test, P = .04 and P = .007, respectively). Univariate analysis showed that baseline glucose consumption of the primary tumor (mean, 0.22; standard deviation [SD], 0.08 μmol/mL/min) and residual MRglu after IC (mean, 0.11; SD, 0.09 μmol/mL/min) provided relevant prognostic information (Table 2): especially the latter was inversely related to survival. Optimal separation between responders and nonresponders was obtained at an MRglu threshold of 0.13 μmol/mL/min after IC (Fig 5).
In a multivariate stepwise analysis, conventional response evaluation (complete or partial response v progressive disease, stable disease, or minor response) significantly improved by adding the absolute level of residual MRglu after completed induction therapy; the final model (CT hazard ratio = 3; 95% CI, 1.3 to 6.9; P = .009; MRglu hazard ratio = 2.1; 95% CI, 1.3 to 3.5; P = .003; Table 3 and Fig 6) showed that PET identified prognostically different strata in the patients who had been considered responsive according to CT.
Prognostic Indicators: After One Cycle of Chemotherapy
After one cycle of induction chemotherapy, the presence of focally enhanced 18FDG uptake in MLN had no predictive value with respect to survival (P = .74). The median fractional change of MRglu in the primary tumor relative to baseline was –37% (range, –15 to –86%). In a preliminary analysis of the first 12 patients, we had found that a 35% decrease of 18FDG uptake discriminated responders from nonresponders with respect to survival (P = .03). Applying this cutoff to the complete data set showed a similar trend (P = .04). However, further univariate analysis showed that, again, absolute levels of residual MRglu (mean, 0.13; SD, 0.07 μmol/mL/min) after one course of IC provided relevant information (P = .01) to the extent that the risk of mortality increased by 2.04 for every 0.1 μmol/mL/min higher MRglu (95% CI, 1.18 to 3.52). Again, the optimal separation between responders and nonresponders was obtained at MRglu 0.13 μmol/mL/min (Fig 7). The pattern of change in 18FDG uptake between the two scans obtained during chemotherapy did not have a significant predictive value with respect to survival (P = .17), even though increasing 18FDG uptake over time tended to be associated with excessive mortality risk.
Simplified PET Quantification
The mean plasma glucose levels with PET scanning (± SD) were 5.3 ± 0.6 mmol/L (range, 3.9 to 6.5 mmol/L). 18FDG uptake measured with simplified methods (SUVBSAg and SKM) showed excellent correlations with MRglu as measured with the Patlak analysis (Pearson correlation coefficients 0.94), as well as in terms of fractional changes after one cycle (r 0.90) and after completed chemotherapy (r 0.90). The nominal values of fractional changes were in the same range for the various techniques: the mean relative changes after one cycle versus baseline were 34% for SKM (SD = 24%) and 33% for SUVBSAg (SD = 23%) and after three cycles versus baseline were 46% (SD = 30%) and 45% (SD = 29%), respectively.
After one cycle of chemotherapy, optimal differentiation between responders and nonresponders was found at SUVBSAg of 140. Patients with an SUVBSAg greater than 140 (n = 18) had a 2-year survival probability of 25% versus 60% for those with lower SUVBSAg (P = .03). After three cycles, the same threshold provided the best discrimination: the 31 patients with SUVBSAg values less than 140 had a 62% probability of surviving 2 years compared with 0% with SUVBSAg 140 (1-year survival estimates, 92% v 55%, respectively; P < .0001).
Again, multivariate analyses including absolute values of SUVBSAg and SKM at baseline, after one cycle, and after three cycles and conventional response measured by CT showed additional value of absolute SUVBSAg and SKM after three cycles in the prediction of survival (Table 4).
DISCUSSION
Validating new biomarkers of response to antitumor therapy is difficult. The current PET literature contains an array of PET methodology and clinical end points. The aims of the present study were first to establish whether the prognostic value of 18FDG PET parameters have any value in selecting patients who will benefit from CMT, and second, to validate the use of simplified PET quantitative methods in terms of clinical relevance. The data obtained in this study show that PET has added value during and after platinum-based therapy and that selected simplified PET methods are suitable for this purpose.
A comprehensive analysis was performed of several potentially relevant prognostic markers that are available from 18FDG PET scans (ie, both at the level of the primary tumor and of MLN metastases). Even though PET surpassed CT with respect to MLN staging after IC, the main finding was that (residual) glucose metabolism of the primary tumor provided the best estimate of prognosis. At the same time, this study provided objective validation of the previously proposed cutoff value of test positivity in the post hoc setting (ie, 50% decrease relative to baseline11,26). Moreover, these data indicated that residual uptake during or after therapy might be an even better surrogate marker of prognosis. Even though the preliminary validation of the cutoff value after one cycle of therapy is promising, objective validation is needed.
In the setting where comparison with CT was feasible (after three cycles), adding this information to the conventional response evaluation improved prognostic stratification. Hence among the 34 patients (74% of the total study group) considered responsive according to standard criteria, PET identified two groups with clearly different prognosis (median survival, 44 months v 18 months). This approach of combining CT and 18FDG PET criteria to document the response has, to the best of our knowledge, not been tested before.
It could be argued that the results of this study were affected by the lack of a single intervention after IC. Unfortunately, this was not possible because of the requirements of the EORTC protocol. On the other hand, it is unlikely that this heterogeneity caused relevant bias, because Taylor et al27 showed no difference in outcome whether patients were treated with surgery or chemoradiotherapy after IC.
Several parameters related to MRglu can be considered for obtaining prognostic strata: fractional change, baseline levels, or residual levels of glucose metabolism. When focusing the PET analysis on the primary tumor, the concept is that the efficacy of systemic therapy to eradicate microscopic distant metastases, the main cause of death in these patients, is reflected by the metabolic changes in the primary tumor. Within the context of CMT of locally advanced NSCLC, residual MRglu measurement seemed to perform best. This parameter obviously combines information about tumor biology before treatment and the impact of therapy as reflected by the change of 18FDG uptake in the primary tumor.28 It has been shown that higher glucose consumption of untreated NSCLC is inversely associated with survival.8,29 However, in other situations (eg, while evaluating the biologic effect of a new drug), relative change might be a better parameter. Presently, the accumulated PET data sets suggest that a drug is unlikely to be effective in the absence of a detectable change in metabolic activity; the reproducibility of 18FDG PET suggests that the minimal detectable change will be approximately 20% to 25%.23,30 Pooling of data from different studies may be easier for fractional changes than for absolute measures, because the former is less sensitive to quantitative PET methodology.21,31,32
MLN status is considered to be an important prognostic factor for survival in patients with locally advanced NSCLC.33 In the present study, visual analysis of 18FDG uptake in MLN had no predictive value early during chemotherapy. It could be argued that a quantitative rather than a visual assessment might have produced other results. However, the present technology does not allow accurate measurement in small lesions that are in the vicinity of blood pool activity. In contrast, after three cycles of IC, visual assessment of MLN status was a significant prognostic factor for survival (P = .04).
The gold standard for measuring MRglu using 18FDG is NLR analysis. Scanning protocol and analysis of this method are relatively complex, and accuracy strongly depends on exact execution of the protocols. Given the consistent high correlation between NLR and Patlak before (r = 0.98), during (r = 0.94) and after IC (r = 0.96), Patlak graphical analysis was used in the present study. For routine clinical practice, simpler semiquantitative methods such as SUVBSAg or SKM would be useful. However, these methods are based on several simplifying assumptions, which may not be valid in repeat (response monitoring) studies and thus may introduce errors in the final analysis.14,34 Results presented here show that correlation between absolute values and fractional change in MRglu with Patlak graphical analysis and the two semiquantitative methods is high and consistent during platinum-based systemic therapy. In future response monitoring studies in this patient group, SUVBSAg (40 to 60 minutes) or SKM (40 to 60 minutes) could be used. It could be suggested that simplified PET methodology is pursued in this setting to obtain further validation of the proposed model. Again, strict replication of PET procedures is necessary to enable meaningful comparisons.
As stated above, we have validated the 50% SUV change threshold proposed by Vansteenkiste et al11 to be used after completion of induction chemotherapy, and this is therefore ready for clinical use. However, if chemoradiotherapy rather than chemotherapy alone is applied, this threshold needs to be confirmed: radiotherapy aims at the same target as the PET evaluation (the primary tumor), and this might modify the association between survival and PET criteria. Regarding PET quantitative technology, with standard chemotherapy and radiotherapy, simplified PET measures (SUV, SKM) can be used instead of the more complex PET quantitative methods, because radiotherapy will not affect tracer pharmacokinetics. We and others have shown that simplified measures are valid for a broad panel of classic chemotherapeutic agents.22,35,36 However, such analysis needs to be repeated for new generations of drugs with uncertain impact on radiotracer biodistribution and kinetics. This can easily be evaluated by performing complex and simplified PET analyses in a limited number of patients on the new drug and comparing these findings with the large database of such evaluations on standard therapies. At the same time, our own proposed threshold (residual FDG uptake) obviously needs external validation in the early response evaluation setting as well as at the end of systemic therapy. After validation of the early response criteria, PET can be used to change patient management. In the setting of induction chemotherapy before definitive local treatment for patients with stage IIIA NSCLC, the main benefit will be to identify patients with poor outcome despite CMT. If such patients are identified shortly after the start of systemic therapy, they should be switched to less toxic local therapy or, preferably, be entered onto trials evaluating alternative therapy strategies.
Finally, this study shows that, despite losing patients because of the technical problems (combination of a dynamic with whole-body scan), a prospective observational study addressing clinical as well as methodologic issues is feasible.
In conclusion, 18FDG PET has additional value over conventional radiologic techniques for monitoring response in locally advanced NSCLC patients treated with CMT. It is feasible to predict response and patient outcome early (after one course of IC) during therapy. In this patient group, semiquantitative methods (SUVBSAg at 40 to 60 minutes and SKM at 40 to 60 minutes) could replace the more complex, quantitative, Patlak graphical analysis.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank Dr Johan Nuyts from the Department of Nuclear Medicine, University Hospital Gasthuisberg, Leuven, Belgium, for providing the semiautomatic program that was used to draw the three-dimensional ROIs.
NOTES
Presented in part at the 40th Annual Meeting of the American Society of Clinical Oncology, June 5-8, 2004, New Orleans, LA.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Therasse P, Arbuck SG, Eisenhauer EA, 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 92:205-216, 2000
Rusch V, Giroux D, Kraut M, et al: Induction chemoradiotherapy and surgical resection for non-small cell lung carcinomas of the superior sulcus (pancoast tumors): Mature results of Southwest Oncology Group trial 9416 (Intergroup trial 0160). Proc Am Soc Clin Oncol 22:634, 2003 (abstr 2548)
Werner-Wasik M, Xiao Y, Pequignot E, et al: Assessment of lung cancer response after nonoperative therapy: Tumor diameter, bidimensional product, and volume—A serial CT scan-based study. Int J Radiat Oncol Biol Phys 51:56-61, 2001
Andre F, Grunenwald D, Pignon JP, et al: Survival of patients with resected N2 non-small-cell lung cancer: Evidence for a subclassification and implications. J Clin Oncol 18:2981-2989, 2000
Eberhardt WE, Albain KS, Pass H, et al: Induction treatment before surgery for non-small cell lung cancer. Lung Cancer 42:S9-14, 2003 (suppl 1)
Weber WA, Petersen V, Schmidt B, 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 21:2651-2657, 2003
Kalff V, Hicks RJ, MacManus MP, et al: Clinical impact of (18)F fluorodeoxyglucose positron emission tomography in patients with non-small-cell lung cancer: A prospective study. J Clin Oncol 19:111-118, 2001
Vansteenkiste JF, Stroobants SG, Dupont PJ, et al: Prognostic importance of the standardized uptake value on (18)F-fluoro-2-deoxy-glucose-positron emission tomography scan in non-small cell lung cancer: An analysis of 125 cases—Leuven Lung Cancer Group. J Clin Oncol 17:3201-3206, 1999
Ryu JS, Choi NC, Fischman AJ, et al: FDG-PET in staging and restaging non-small cell lung cancer after neoadjuvant chemoradiotherapy: Correlation with histopathology. Lung Cancer 35:179-187, 2002
Cerfolio RJ, Ojha B, Mukherjee S, et al: Positron emission tomography scanning with 2-fluoro-2-deoxy-d-glucose as a predictor of response of neoadjuvant treatment for non-small cell carcinoma. J Thorac Cardiovasc Surg 125:938-944, 2003
Vansteenkiste JF, Stroobants SG, De Leyn PR, et al: Potential use of FDG-PET scan after induction chemotherapy in surgically staged IIIa-N2 non-small-cell lung cancer: A prospective pilot study—The Leuven Lung Cancer Group. Ann Oncol 9:1193-1198, 1998
Mac Manus MP, Hicks RJ, Matthews JP, et al: Positron emission tomography is superior to computed tomography scanning for response-assessment after radical radiotherapy or chemoradiotherapy in patients with non-small-cell lung cancer. J Clin Oncol 21:1285-1292, 2003
Patz EF, Connolly J, Herndon J: Prognostic value of thoracic FDG PET imaging after treatment for non-small cell lung cancer. AJR Am J Roentgenol 174:769-774, 2000
Hoekstra CJ, Paglianiti I, Hoekstra OS, et al: Monitoring response to therapy in cancer using [18F]-2-fluoro-2-deoxy-D-glucose and positron emission tomography: An overview of different analytical methods. Eur J Nucl Med 27:731-743, 2000
Clinical practice guidelines for the treatment of unresectable non-small-cell lung cancer: Adopted on May 16, 1997 by the American Society of Clinical Oncology. J Clin Oncol 15:2996-3018, 1997
Lorent N, Vansteenkiste J, De Leyn PR, et al: Long-term survival of surgically staged IIIA-N2 non-small cell lung cancer treated with surgical combined modality approach: Analysis of a 7-year experience. Ann Oncol 15:1645-1653, 2004
Hoekstra CJ, Stroobants SG, Hoekstra OS, et al: The value of [18F]fluoro-2-deoxy-D-glucose positron emission tomography in the selection of patients with stage IIIA-N2 non-small cell lung cancer for combined modality treatment. Lung Cancer 39:151-157, 2003
Boellaard R, van Balen SC, Lammertsma AA: Optimization of attenuation correction for PET studies of thorax and pelvis using count-ased transmission scans. J Nucl Med 42:196, 2001
Hoekstra CJ, Hoekstra OS, Lammertsma AA: On the use of image-derived input functions in oncological fluorine-18 fluorodeoxyglucose positron emission tomography studies. Eur J Nucl Med 26:1489-1492, 1999
van der Weerdt AP, Klein LJ, Boellaard R, et al: Image-derived input functions for determination of MRGlu in cardiac (18)F-FDG PET scans. J Nucl Med 42:1622-1629, 2001
Boellaard R, Krak NC, Hoekstra OS, et al: Effects of noise, image resolution, and ROI definition on the accuracy of standard uptake values: A simulation study. J Nucl Med 45:1519-1527, 2004
Boellaard R, van Lingen A, Lammertsma AA: Experimental and clinical evaluation of iterative reconstruction (osem) in dynamic PET: Quantitative characteristics and effects on kinetic modeling. J Nucl Med 42:808-817, 2001
Hoekstra CJ, Hoekstra OS, Stroobants SG, et al: Methods to monitor response to chemotherapy in non-small cell lung cancer with 18F-FDG PET. J Nucl Med 43:1304-1309, 2002
Hunter GJ, Hamberg LM, Alpert NM, et al: Simplified measurement of deoxyglucose utilization rate. J Nucl Med 37:950-955, 1996
Akaike H: A new look at the statistical identification. IEEE Trans Automat Contr 19:716-723, 1978
Port JL, Kent MS, Korst RJ, et al: Positron emission tomography scanning poorly predicts response to preoperative chemotherapy in non-small cell lung cancer. Ann Thorac Surg 77:254-259, 2004
Taylor NA, Liao ZX, Cox JD, et al: Equivalent outcome of patients with clinical Stage IIIA non-small-cell lung cancer treated with concurrent chemoradiation compared with induction chemotherapy followed by surgical resection. Int J Radiat Oncol Biol Phys 58:204-212, 2004
Downey RJ, Akhurst T, Ilson D, et al: Whole body 18FDG-PET and the response of esophageal cancer to induction therapy: Results of a prospective trial. J Clin Oncol 21:428-432, 2003
Downey RJ, Akhurst T, Gonen M, et al: Preoperative F-18 fluorodeoxyglucose-positron emission tomography maximal standardized uptake value predicts survival after lung cancer resection. J Clin Oncol 22:3255-3260, 2004
Weber WA, Ziegler SI, Thodtmann R, et al: Reproducibility of metabolic measurements in malignant tumors using FDG PET. J Nucl Med 40:1771-1777, 1999
Stahl A, Ott K, Schwaiger M, Weber WA: Comparison of different SUV-based methods for monitoring cytotoxic therapy with FDG PET. Eur J Nucl Med Mol Imaging 31:1471-1478, 2004
Krak NC, Boellaard R, Hoekstra OS, et al: Effects of ROI definition and reconstruction method on quantitative outcome and applicability in a response monitoring trial. Eur J Nucl Med Mol Imaging 32:294-301, 2005
Vansteenkiste JF, De Leyn PR, Deneffe GJ, et al: Clinical prognostic factors in surgically treated stage IIIA-N2 non-small cell lung cancer: Analysis of the literature. Lung Cancer 19:3-13, 1998
Keyes JW Jr: SUV: Standard uptake or silly useless value? J Nucl Med 36:1836-1839, 1995
Krak NC, van der Hoeven JJM, Hoekstra OS, et al: Measuring [(18)F]FDG uptake in breast cancer during chemotherapy: Comparison of analytical methods. Eur J Nucl Med Mol Imaging 30:674-681, 2003
Kroep JR, van Groeningen CJ, Cuesta MA, et al: Positron emission tomography using 2-deoxy-2-[18F]-fluoro-D-glucose for response monitoring in locally advanced gastroesophageal cancer: A comparison of different analytical methods. Mol Imaging Biol 5:337-346, 2003(Corneline J. Hoekstra, Si)
Department of Pulmonary Medicine, Jeroen Bosch Hospital, 's-Hertogenbosch
Department of Pulmonary Medicine, St Antonius Hospital, Nieuwegein
Department of Thoracic Oncology, Cancer Institute/Antoni van Leeuwenhoek Hospital
Comprehensive Cancer Center, Amsterdam, the Netherlands
Departments of Nuclear Medicine and Pulmonary Medicine, University Hospital Gasthuisberg, Leuven, Belgium
ABSTRACT
PURPOSE: The objective of this study was to determine the accuracy of (early) response measurements using [18F]-2-fluoro-2-deoxy-D-glucose positron emission tomography (18FDG PET) with respect to survival of patients with stage IIIA-N2 non–small-cell lung cancer (NSCLC) undergoing induction chemotherapy (IC), with a comparative analysis of PET methods.
PATIENTS AND METHODS: In a prospective multicenter study, PET was performed in patients before IC and after one and three cycles. Computed tomography (CT) was performed before and after IC. Glucose consumption (metabolic rate of glucose [MRglu]) was measured using Patlak graphical analysis and correlated with simplified methods. Mediastinal lymph node (MLN) status was assessed visually. Cox proportional hazards analysis was used to determine the prognostic relevance of CT and PET measures of response with respect to survival.
RESULTS: Complete PET data sets were available in 47 patients. Median survival was 21 months. MLN status after IC by PET predicted survival (hazard ratio [HR], 2.33; 95% CI, 1.04 to 5.22; P = .04) in contrast with CT (HR, 1.87; 95% CI, 0.81 to 4.30; P = .14). Residual MRglu after IC proved to be the best prognostic factor (HR, 1.95; 95% CI, 1.28 to 2.97; P = .002). Multivariate stepwise analysis showed that PET identified prognostically different strata in patients considered responsive according to CT. Residual MRglu after one cycle selected patients with different outcomes (HR, 2.04; 95% CI, 1.18 to 3.52; P = .01). Simplified quantitative 18FDG PET methods were correlated with Patlak graphical analysis during and after therapy (r 0.90).
CONCLUSION: 18FDG PET has additional value over CT in monitoring response to IC in patients with stage IIIA-N2 NSCLC, and it seems feasible to predict survival early during IC. Simple semiquantitative and complex PET methods perform equally well.
INTRODUCTION
At present, monitoring response to induction chemotherapy (IC) in non–small-cell lung cancer (NSCLC) is a post hoc approach; in other words, at the end of treatment, volumetric changes of known tumor sites are measured.1 In locally advanced NSCLC (stage IIIA-N2), the clinical utility of standard computed tomography (CT) criteria to document response to IC is limited.2,3 Although retrospective studies have suggested that radiologic response to IC carries prognostic information with respect to survival,4 in the majority of patients, information with indeterminate prognostic relevance is obtained, as typical response rates (mostly partial) to IC are approximately 70%.5 A more reliable stratification of patients subjected to combined-modality therapy (CMT) could have major clinical implications, especially if prognostically relevant information would be available at an early stage of induction treatment.
Positron emission tomography (PET) assesses the effect of therapy at a metabolic rather than at an anatomic level. [18F]-2-fluoro-2-deoxy-D-glucose (18FDG) has high affinity for NSCLC, and as such, whole-body PET improves the selection of patients eligible for CMT.6-8
There are some data to suggest that changes of glucose metabolism as measured with 18FDG PET after completed therapy reflect therapeutic effectiveness more accurately than volumetric measures.6,9-13 However, assessment of the value of 18FDG PET for monitoring response is complicated by heterogeneity of published data with respect to the applied methods of PET quantification (comprising an array of visual and [semi-] quantitative measures14), the primary targets of PET evaluation (primary tumor, lymph node status), and the clinical end points (histology, survival). As a result, objective validation of proposed thresholds of responsiveness for PET is lacking. It is unclear whether and how application of 18FDG PET improves the present standard of radiologic response classification. Finally, the potential of therapy evaluation with PET at an earlier and perhaps clinically more relevant stage of NSCLC treatment has only been demonstrated for stage IV NSCLC.6
The objective of the present study was to investigate the accuracy of (early) response evaluation using 18FDG PET to predict survival of patients undergoing CMT for stage IIIA-N2 NSCLC using a study design, which allowed for comparative analysis of quantitative PET methods at various levels of complexity.
PATIENTS AND METHODS
Between January 1998 and May 2001, this multicenter study was open for accrual. Before PET, patients had been staged as pIIIA-N2 NSCLC and were considered candidates for CMT. Initial staging was performed according to the guidelines of the American Society of Clinical Oncology.15 Patients with insulin-dependent diabetes mellitus were excluded. PET scans were performed at the clinical PET center of the Vrije University Medical Center, Amsterdam, the Netherlands, or the University Hospital Gasthuisberg, Leuven, Belgium. All patients received three cycles of platinum-based IC, usually containing a third-generation cytotoxic agent (Fig 1). In the Netherlands, patients without progression of disease were subsequently randomly assigned to either surgery or radiotherapy according to study protocol 08941 by the European Organization for Research and Treatment of Cancer (EORTC; available at www.eortc.be). In Belgium, patients were treated in a prospective protocol of vindesine, ifosfamide, and platinum induction followed by attempted complete resection,16 whereas radical radiotherapy was applied only in case of unacceptable surgical risk because of cardiopulmonary limitations, when the disease was considered irresectable postinduction by the attending thoracic surgeon, or on patient's request. The medical ethics committees of all participating institutes approved the study protocol. All patients gave informed consent before inclusion according to local medical ethical committee regulations.
PET scans were performed after conventional staging was completed. Dynamic PET scans were performed before (1 to 2 weeks) the start of IC, before the second cycle (1 to 5 days), and within 3 to 4 weeks after completion of the third cycle of IC, but before the start of locoregional therapy. In addition, we performed whole-body scans before IC and after the third cycle. Clinicians were blinded to the results of the dynamic PET scans but were informed about unexpected metastatic findings on the whole-body scans, in the best interest of the patient. In the context of patient care, upstaging by PET had to be confirmed with conventional imaging or pathology before altering therapy.17
Radiotherapy
The aim of radiotherapy was to irradiate the primary tumor and areas of radiologic involvement with at least 2-cm margins to a dose of 60.0 to 62.5 Gy, using daily fractions of 2 Gy, once per day, 5 days a week. The lymphatic drainage areas of the mediastinum, which were not abnormal on CT, received a prophylactic dose of 40 to 45 Gy in daily fractions of 2 Gy.
Surgery
Thoracotomy with systematic nodal dissection was performed with the aim of complete resection of the primary tumor, ipsilateral hilar nodes, and mediastinal lymph nodes. Even in case of extensive nodal involvement, complete resection was attempted. Complete resection was defined as macroscopic and microscopic free tumor margins. The highest mediastinal node had to be free of tumor. In case of unsuspected metastasis, or when it was clear that resection would be macroscopically (and proven by frozen section) incomplete, only systematic mediastinal lymphadenectomy for staging purposes was performed, provided that the per-operative condition of the patient allowed this.
PET Acquisition
Scans were performed using state-of-the-art PET scanners (ECAT EXACT HR+; Siemens/CTI, Knoxville, TN). These scanners have an axial field of view of 15 cm, divided into 63 contiguous planes. Patients were positioned supine on the scanner bed with the tumor in the center of the axial field of view of the scanner. Patients had fasted for at least 6 hours before scanning.
Acquisition started with a 10- to 15-minute transmission scan to correct for photon attenuation,18 followed by a bolus injection of 370 MBq of 18FDG in 5 mL of saline through an injector (Medrad International, Maastricht, the Netherlands) at a rate of 0.8 mL/sec, after which the line was flushed with 42 mL of saline (2.0 mL/sec). Simultaneously with the injection of 18FDG, a dynamic emission scan (in two-dimensional mode) was started with a total duration of 60 minutes with variable frame length (6 x 5, 6 x 10, 3 x 20, 5 x 30, 5 x 60, 8 x 150, and 6 x 300 seconds). All dynamic scan data were corrected for dead time, decay, scatter, randoms, and photon attenuation and were reconstructed as 128 x 128 matrices using filtered back projection (FBP) with a Hanning filter (cutoff, 0.5 cycles/pixel). This resulted in a transaxial spatial resolution of around 7-mm full width at half maximum. In addition, three venous blood samples were drawn at 35, 45, and 55 minutes after 18FDG injection as quality control for the image-derived input function19,20 and for plasma glucose measurement (hexokinase method, Hitachi 747; Boeringer Mannheim, Mannheim, Germany).
Data Analysis
An experienced radiologist blinded to patient data assessed tumor response. Response was based on the comparison of two-dimensional measurable lesions on a chest spiral CT scan obtained before and after three cycles of IC using WHO criteria.15 Persistent N2 disease after IC was defined on CT as persistence of enlarged MLN (> 1 cm in the short axis).
Quantitative PET data analysis was performed without any knowledge of clinical outcome. Using 18FDG PET data, three-dimensional regions of interest (ROIs) were defined semiautomatically over the tumor using a threshold of 50% of the maximum pixel value within the tumor after background correction (using a manually drawn ROI over normal tissue next to the tumor).21,32 The (local) background correction is applied to obtain realistic volumes of interest (VOI) in case of low contrast between tumor and background, and should, in theory, most closely correspond to actual tumor volume. In fact, it provides a correction for the limited resolution of PET images. In short, first the isocontour at 70% of tumor maximum is defined automatically as a first guess of tumor extent. Next, a three-dimensional shell of 1 voxel thickness at approximately 20 mm (three times the image resolution) distance from the outer voxels of the 70% VOI was defined. The average voxel value of the voxels within this shell is used as an estimate of average local background. Finally, background corrected VOIs are generated by taking the 50% value of tumor maximum minus this background value. For this purpose the last three frames of the sinograms (45 of 60 minutes postinjection) were summed and reconstructed using ordered subset expectation maximization with two iterations and 12 subsets followed by postsmoothing of the reconstructed image using a 5-mm full width at half maximum Gaussian filter was used to obtain the same resolution as FBP data.22 In addition, using FBP data, ROIs were defined manually over the aortic arch, left ventricle, and left atrium to obtain an image-derived input function, as described previously.19 MLN metastases were analyzed visually; focally enhanced activity above mediastinal background was considered positive.
Initially, response analysis of the primary tumor was performed using nonlinear regression (NLR), using the standard two-tissue compartment model with three (3k) rate constants, a blood volume component, and an image-derived input function. A known disadvantage of NLR is its sensitivity to noise, which will increase for low values of glucose consumption (MRglu) in a lesion (ie, after good response). In a previous study,23 a high correlation between NLR and Patlak graphical analysis at 10 to 60 minutes postinjection (a method less susceptible to noise in the data) was found. A correlation analysis for the present data showed improvement in this correlation (r = 0.98 to 0.99) when leaving out a few obvious outliers in the NLR data. On the basis of this observation, Patlak graphical analysis was used to analyze the study data. In addition, two simplified methods were used23: the standard uptake value (SUV) at 40 to 60 minutes postinjection (corrected for body-surface area and plasma glucose; SUVBSAg), and the simplified kinetic method (SKM) as described by Hunter et al24 at 40 to 60 minutes postinjection. All methods were applied to the same data set. A more detailed description of these methods can be found elsewhere.14 The lumped constant, which accounts for differences between glucose and 18FDG, was set to 1 and assumed to be constant over time, as no studies on the actual value of the lumped constant in tumors outside the CNS have been reported.14
Follow-Up
Every 2 months after definitive locoregional therapy, patients underwent blood tests, physical examination, and chest radiography. When tumor relapse was suspected, additional tests were performed.
Statistics
The correlation of percentage change in MRglu over time with Patlak was assessed. Cox proportional hazards regression analysis was used to assess the prognostic value of the baseline and absolute uptake after one and three courses, respectively, and the change relative to baseline MRglu and after one and three courses with respect to survival time.
Results are presented by hazard ratios with 95% CIs. Basic assumptions of linearity and additivity were checked. The value of MLN status as assessed by CT and 18FDG PET were analyzed using Kaplan and Meier plots and the log-rank test (P < .05). Finally, models were constructed of variables related to baseline and change in uptake after one cycle, and these were compared by means of the Akaike Information Criterion.25 Cut points were established by means of exploring martingale residual plots and the minimum P value approach.
MRglu results at baseline and after one and three courses of IC by Patlak analysis were then correlated with SUVBSAg and SKM results.
RESULTS
Seventy-nine patients were included in the study. After the first PET scan, but before IC, 23 patients were excluded because of confirmed upstaging (stage IIIB/IV, n = 12), technical problems (n = 9), or patient refusal (n = 2). After the second PET scan (after one IC cycle), another nine patients were excluded because of technical problems (n = 4), exclusion from the EORTC protocol (n = 4), or refusal to undergo a third PET scan (n = 1; Fig 2).
Therefore, complete PET data sets were available in 47 patients (Table 1) to be used for analysis of prognostic value. After induction chemotherapy, 25 patients were treated with surgery, 15 patients were treated with radical radiotherapy, and six patients were treated with radiotherapy with palliative intent because of progression based on CT criteria during IC. In the remaining patient, no locoregional therapy was given, as response to IC was unclear and his physical condition deteriorated. Resection was radical in 21 patients. Irradical resection as a result of persisting MLN involvement was found in three patients, and in one patient, the primary tumor could not be radically removed.
Median follow-up of all patients was 28 months, with a minimum of 10 months in surviving patients. During the first 2 years of follow-up, 24 patients died of disease, and recurrent disease was established in five other patients. Median survival was 21 months. In 10 patients, PET had suggested N3 (n = 2) or distant (n = 8) metastatic disease, which could not be confirmed clinically. According to the study protocol, these patients were treated with multimodality therapy, and their overall survival was similar to that of the remaining patients (P = .74).
Prognostic Indicators: After Induction Therapy
At the end of IC, N stage at CT was not predictive of survival (log-rank test, P = .14), but focally enhanced MLN 18FDG uptake was associated with a two-fold higher risk of mortality (log-rank test, P = .04; Fig 3). In the operated subset (n = 25), sensitivity and specificity of 18FDG PET after IC for MLN (N2) involvement were 50% (95% CI, 19% to 81%) and 71% (95% CI, 42% to 92%), respectively (corresponding to positive and negative predictive values of 66% [95% CI, 21% to 86%] and 67% [95% CI, 38% to 88%], respectively).
Using standard WHO criteria, the majority of patients (n = 26) had a partial response after completed IC, six patients had a minor response, seven patients had stable disease, five patients had progressive disease, and two patients showed complete response at CT. In one patient, response could not be classified as no CT scan was performed before the start of IC.
The median fractional change of metabolic rate of glucose (MRglu) with respect to baseline was –48% (range, –28 to –97%; Fig 4). The previously defined PET criterion for response (50% decrease of [18F]FDG uptake after three cycles v baseline11) either expressed as MRglu or SUVBSAg predicted survival significantly (log-rank test, P = .04 and P = .007, respectively). Univariate analysis showed that baseline glucose consumption of the primary tumor (mean, 0.22; standard deviation [SD], 0.08 μmol/mL/min) and residual MRglu after IC (mean, 0.11; SD, 0.09 μmol/mL/min) provided relevant prognostic information (Table 2): especially the latter was inversely related to survival. Optimal separation between responders and nonresponders was obtained at an MRglu threshold of 0.13 μmol/mL/min after IC (Fig 5).
In a multivariate stepwise analysis, conventional response evaluation (complete or partial response v progressive disease, stable disease, or minor response) significantly improved by adding the absolute level of residual MRglu after completed induction therapy; the final model (CT hazard ratio = 3; 95% CI, 1.3 to 6.9; P = .009; MRglu hazard ratio = 2.1; 95% CI, 1.3 to 3.5; P = .003; Table 3 and Fig 6) showed that PET identified prognostically different strata in the patients who had been considered responsive according to CT.
Prognostic Indicators: After One Cycle of Chemotherapy
After one cycle of induction chemotherapy, the presence of focally enhanced 18FDG uptake in MLN had no predictive value with respect to survival (P = .74). The median fractional change of MRglu in the primary tumor relative to baseline was –37% (range, –15 to –86%). In a preliminary analysis of the first 12 patients, we had found that a 35% decrease of 18FDG uptake discriminated responders from nonresponders with respect to survival (P = .03). Applying this cutoff to the complete data set showed a similar trend (P = .04). However, further univariate analysis showed that, again, absolute levels of residual MRglu (mean, 0.13; SD, 0.07 μmol/mL/min) after one course of IC provided relevant information (P = .01) to the extent that the risk of mortality increased by 2.04 for every 0.1 μmol/mL/min higher MRglu (95% CI, 1.18 to 3.52). Again, the optimal separation between responders and nonresponders was obtained at MRglu 0.13 μmol/mL/min (Fig 7). The pattern of change in 18FDG uptake between the two scans obtained during chemotherapy did not have a significant predictive value with respect to survival (P = .17), even though increasing 18FDG uptake over time tended to be associated with excessive mortality risk.
Simplified PET Quantification
The mean plasma glucose levels with PET scanning (± SD) were 5.3 ± 0.6 mmol/L (range, 3.9 to 6.5 mmol/L). 18FDG uptake measured with simplified methods (SUVBSAg and SKM) showed excellent correlations with MRglu as measured with the Patlak analysis (Pearson correlation coefficients 0.94), as well as in terms of fractional changes after one cycle (r 0.90) and after completed chemotherapy (r 0.90). The nominal values of fractional changes were in the same range for the various techniques: the mean relative changes after one cycle versus baseline were 34% for SKM (SD = 24%) and 33% for SUVBSAg (SD = 23%) and after three cycles versus baseline were 46% (SD = 30%) and 45% (SD = 29%), respectively.
After one cycle of chemotherapy, optimal differentiation between responders and nonresponders was found at SUVBSAg of 140. Patients with an SUVBSAg greater than 140 (n = 18) had a 2-year survival probability of 25% versus 60% for those with lower SUVBSAg (P = .03). After three cycles, the same threshold provided the best discrimination: the 31 patients with SUVBSAg values less than 140 had a 62% probability of surviving 2 years compared with 0% with SUVBSAg 140 (1-year survival estimates, 92% v 55%, respectively; P < .0001).
Again, multivariate analyses including absolute values of SUVBSAg and SKM at baseline, after one cycle, and after three cycles and conventional response measured by CT showed additional value of absolute SUVBSAg and SKM after three cycles in the prediction of survival (Table 4).
DISCUSSION
Validating new biomarkers of response to antitumor therapy is difficult. The current PET literature contains an array of PET methodology and clinical end points. The aims of the present study were first to establish whether the prognostic value of 18FDG PET parameters have any value in selecting patients who will benefit from CMT, and second, to validate the use of simplified PET quantitative methods in terms of clinical relevance. The data obtained in this study show that PET has added value during and after platinum-based therapy and that selected simplified PET methods are suitable for this purpose.
A comprehensive analysis was performed of several potentially relevant prognostic markers that are available from 18FDG PET scans (ie, both at the level of the primary tumor and of MLN metastases). Even though PET surpassed CT with respect to MLN staging after IC, the main finding was that (residual) glucose metabolism of the primary tumor provided the best estimate of prognosis. At the same time, this study provided objective validation of the previously proposed cutoff value of test positivity in the post hoc setting (ie, 50% decrease relative to baseline11,26). Moreover, these data indicated that residual uptake during or after therapy might be an even better surrogate marker of prognosis. Even though the preliminary validation of the cutoff value after one cycle of therapy is promising, objective validation is needed.
In the setting where comparison with CT was feasible (after three cycles), adding this information to the conventional response evaluation improved prognostic stratification. Hence among the 34 patients (74% of the total study group) considered responsive according to standard criteria, PET identified two groups with clearly different prognosis (median survival, 44 months v 18 months). This approach of combining CT and 18FDG PET criteria to document the response has, to the best of our knowledge, not been tested before.
It could be argued that the results of this study were affected by the lack of a single intervention after IC. Unfortunately, this was not possible because of the requirements of the EORTC protocol. On the other hand, it is unlikely that this heterogeneity caused relevant bias, because Taylor et al27 showed no difference in outcome whether patients were treated with surgery or chemoradiotherapy after IC.
Several parameters related to MRglu can be considered for obtaining prognostic strata: fractional change, baseline levels, or residual levels of glucose metabolism. When focusing the PET analysis on the primary tumor, the concept is that the efficacy of systemic therapy to eradicate microscopic distant metastases, the main cause of death in these patients, is reflected by the metabolic changes in the primary tumor. Within the context of CMT of locally advanced NSCLC, residual MRglu measurement seemed to perform best. This parameter obviously combines information about tumor biology before treatment and the impact of therapy as reflected by the change of 18FDG uptake in the primary tumor.28 It has been shown that higher glucose consumption of untreated NSCLC is inversely associated with survival.8,29 However, in other situations (eg, while evaluating the biologic effect of a new drug), relative change might be a better parameter. Presently, the accumulated PET data sets suggest that a drug is unlikely to be effective in the absence of a detectable change in metabolic activity; the reproducibility of 18FDG PET suggests that the minimal detectable change will be approximately 20% to 25%.23,30 Pooling of data from different studies may be easier for fractional changes than for absolute measures, because the former is less sensitive to quantitative PET methodology.21,31,32
MLN status is considered to be an important prognostic factor for survival in patients with locally advanced NSCLC.33 In the present study, visual analysis of 18FDG uptake in MLN had no predictive value early during chemotherapy. It could be argued that a quantitative rather than a visual assessment might have produced other results. However, the present technology does not allow accurate measurement in small lesions that are in the vicinity of blood pool activity. In contrast, after three cycles of IC, visual assessment of MLN status was a significant prognostic factor for survival (P = .04).
The gold standard for measuring MRglu using 18FDG is NLR analysis. Scanning protocol and analysis of this method are relatively complex, and accuracy strongly depends on exact execution of the protocols. Given the consistent high correlation between NLR and Patlak before (r = 0.98), during (r = 0.94) and after IC (r = 0.96), Patlak graphical analysis was used in the present study. For routine clinical practice, simpler semiquantitative methods such as SUVBSAg or SKM would be useful. However, these methods are based on several simplifying assumptions, which may not be valid in repeat (response monitoring) studies and thus may introduce errors in the final analysis.14,34 Results presented here show that correlation between absolute values and fractional change in MRglu with Patlak graphical analysis and the two semiquantitative methods is high and consistent during platinum-based systemic therapy. In future response monitoring studies in this patient group, SUVBSAg (40 to 60 minutes) or SKM (40 to 60 minutes) could be used. It could be suggested that simplified PET methodology is pursued in this setting to obtain further validation of the proposed model. Again, strict replication of PET procedures is necessary to enable meaningful comparisons.
As stated above, we have validated the 50% SUV change threshold proposed by Vansteenkiste et al11 to be used after completion of induction chemotherapy, and this is therefore ready for clinical use. However, if chemoradiotherapy rather than chemotherapy alone is applied, this threshold needs to be confirmed: radiotherapy aims at the same target as the PET evaluation (the primary tumor), and this might modify the association between survival and PET criteria. Regarding PET quantitative technology, with standard chemotherapy and radiotherapy, simplified PET measures (SUV, SKM) can be used instead of the more complex PET quantitative methods, because radiotherapy will not affect tracer pharmacokinetics. We and others have shown that simplified measures are valid for a broad panel of classic chemotherapeutic agents.22,35,36 However, such analysis needs to be repeated for new generations of drugs with uncertain impact on radiotracer biodistribution and kinetics. This can easily be evaluated by performing complex and simplified PET analyses in a limited number of patients on the new drug and comparing these findings with the large database of such evaluations on standard therapies. At the same time, our own proposed threshold (residual FDG uptake) obviously needs external validation in the early response evaluation setting as well as at the end of systemic therapy. After validation of the early response criteria, PET can be used to change patient management. In the setting of induction chemotherapy before definitive local treatment for patients with stage IIIA NSCLC, the main benefit will be to identify patients with poor outcome despite CMT. If such patients are identified shortly after the start of systemic therapy, they should be switched to less toxic local therapy or, preferably, be entered onto trials evaluating alternative therapy strategies.
Finally, this study shows that, despite losing patients because of the technical problems (combination of a dynamic with whole-body scan), a prospective observational study addressing clinical as well as methodologic issues is feasible.
In conclusion, 18FDG PET has additional value over conventional radiologic techniques for monitoring response in locally advanced NSCLC patients treated with CMT. It is feasible to predict response and patient outcome early (after one course of IC) during therapy. In this patient group, semiquantitative methods (SUVBSAg at 40 to 60 minutes and SKM at 40 to 60 minutes) could replace the more complex, quantitative, Patlak graphical analysis.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank Dr Johan Nuyts from the Department of Nuclear Medicine, University Hospital Gasthuisberg, Leuven, Belgium, for providing the semiautomatic program that was used to draw the three-dimensional ROIs.
NOTES
Presented in part at the 40th Annual Meeting of the American Society of Clinical Oncology, June 5-8, 2004, New Orleans, LA.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Therasse P, Arbuck SG, Eisenhauer EA, 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 92:205-216, 2000
Rusch V, Giroux D, Kraut M, et al: Induction chemoradiotherapy and surgical resection for non-small cell lung carcinomas of the superior sulcus (pancoast tumors): Mature results of Southwest Oncology Group trial 9416 (Intergroup trial 0160). Proc Am Soc Clin Oncol 22:634, 2003 (abstr 2548)
Werner-Wasik M, Xiao Y, Pequignot E, et al: Assessment of lung cancer response after nonoperative therapy: Tumor diameter, bidimensional product, and volume—A serial CT scan-based study. Int J Radiat Oncol Biol Phys 51:56-61, 2001
Andre F, Grunenwald D, Pignon JP, et al: Survival of patients with resected N2 non-small-cell lung cancer: Evidence for a subclassification and implications. J Clin Oncol 18:2981-2989, 2000
Eberhardt WE, Albain KS, Pass H, et al: Induction treatment before surgery for non-small cell lung cancer. Lung Cancer 42:S9-14, 2003 (suppl 1)
Weber WA, Petersen V, Schmidt B, 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 21:2651-2657, 2003
Kalff V, Hicks RJ, MacManus MP, et al: Clinical impact of (18)F fluorodeoxyglucose positron emission tomography in patients with non-small-cell lung cancer: A prospective study. J Clin Oncol 19:111-118, 2001
Vansteenkiste JF, Stroobants SG, Dupont PJ, et al: Prognostic importance of the standardized uptake value on (18)F-fluoro-2-deoxy-glucose-positron emission tomography scan in non-small cell lung cancer: An analysis of 125 cases—Leuven Lung Cancer Group. J Clin Oncol 17:3201-3206, 1999
Ryu JS, Choi NC, Fischman AJ, et al: FDG-PET in staging and restaging non-small cell lung cancer after neoadjuvant chemoradiotherapy: Correlation with histopathology. Lung Cancer 35:179-187, 2002
Cerfolio RJ, Ojha B, Mukherjee S, et al: Positron emission tomography scanning with 2-fluoro-2-deoxy-d-glucose as a predictor of response of neoadjuvant treatment for non-small cell carcinoma. J Thorac Cardiovasc Surg 125:938-944, 2003
Vansteenkiste JF, Stroobants SG, De Leyn PR, et al: Potential use of FDG-PET scan after induction chemotherapy in surgically staged IIIa-N2 non-small-cell lung cancer: A prospective pilot study—The Leuven Lung Cancer Group. Ann Oncol 9:1193-1198, 1998
Mac Manus MP, Hicks RJ, Matthews JP, et al: Positron emission tomography is superior to computed tomography scanning for response-assessment after radical radiotherapy or chemoradiotherapy in patients with non-small-cell lung cancer. J Clin Oncol 21:1285-1292, 2003
Patz EF, Connolly J, Herndon J: Prognostic value of thoracic FDG PET imaging after treatment for non-small cell lung cancer. AJR Am J Roentgenol 174:769-774, 2000
Hoekstra CJ, Paglianiti I, Hoekstra OS, et al: Monitoring response to therapy in cancer using [18F]-2-fluoro-2-deoxy-D-glucose and positron emission tomography: An overview of different analytical methods. Eur J Nucl Med 27:731-743, 2000
Clinical practice guidelines for the treatment of unresectable non-small-cell lung cancer: Adopted on May 16, 1997 by the American Society of Clinical Oncology. J Clin Oncol 15:2996-3018, 1997
Lorent N, Vansteenkiste J, De Leyn PR, et al: Long-term survival of surgically staged IIIA-N2 non-small cell lung cancer treated with surgical combined modality approach: Analysis of a 7-year experience. Ann Oncol 15:1645-1653, 2004
Hoekstra CJ, Stroobants SG, Hoekstra OS, et al: The value of [18F]fluoro-2-deoxy-D-glucose positron emission tomography in the selection of patients with stage IIIA-N2 non-small cell lung cancer for combined modality treatment. Lung Cancer 39:151-157, 2003
Boellaard R, van Balen SC, Lammertsma AA: Optimization of attenuation correction for PET studies of thorax and pelvis using count-ased transmission scans. J Nucl Med 42:196, 2001
Hoekstra CJ, Hoekstra OS, Lammertsma AA: On the use of image-derived input functions in oncological fluorine-18 fluorodeoxyglucose positron emission tomography studies. Eur J Nucl Med 26:1489-1492, 1999
van der Weerdt AP, Klein LJ, Boellaard R, et al: Image-derived input functions for determination of MRGlu in cardiac (18)F-FDG PET scans. J Nucl Med 42:1622-1629, 2001
Boellaard R, Krak NC, Hoekstra OS, et al: Effects of noise, image resolution, and ROI definition on the accuracy of standard uptake values: A simulation study. J Nucl Med 45:1519-1527, 2004
Boellaard R, van Lingen A, Lammertsma AA: Experimental and clinical evaluation of iterative reconstruction (osem) in dynamic PET: Quantitative characteristics and effects on kinetic modeling. J Nucl Med 42:808-817, 2001
Hoekstra CJ, Hoekstra OS, Stroobants SG, et al: Methods to monitor response to chemotherapy in non-small cell lung cancer with 18F-FDG PET. J Nucl Med 43:1304-1309, 2002
Hunter GJ, Hamberg LM, Alpert NM, et al: Simplified measurement of deoxyglucose utilization rate. J Nucl Med 37:950-955, 1996
Akaike H: A new look at the statistical identification. IEEE Trans Automat Contr 19:716-723, 1978
Port JL, Kent MS, Korst RJ, et al: Positron emission tomography scanning poorly predicts response to preoperative chemotherapy in non-small cell lung cancer. Ann Thorac Surg 77:254-259, 2004
Taylor NA, Liao ZX, Cox JD, et al: Equivalent outcome of patients with clinical Stage IIIA non-small-cell lung cancer treated with concurrent chemoradiation compared with induction chemotherapy followed by surgical resection. Int J Radiat Oncol Biol Phys 58:204-212, 2004
Downey RJ, Akhurst T, Ilson D, et al: Whole body 18FDG-PET and the response of esophageal cancer to induction therapy: Results of a prospective trial. J Clin Oncol 21:428-432, 2003
Downey RJ, Akhurst T, Gonen M, et al: Preoperative F-18 fluorodeoxyglucose-positron emission tomography maximal standardized uptake value predicts survival after lung cancer resection. J Clin Oncol 22:3255-3260, 2004
Weber WA, Ziegler SI, Thodtmann R, et al: Reproducibility of metabolic measurements in malignant tumors using FDG PET. J Nucl Med 40:1771-1777, 1999
Stahl A, Ott K, Schwaiger M, Weber WA: Comparison of different SUV-based methods for monitoring cytotoxic therapy with FDG PET. Eur J Nucl Med Mol Imaging 31:1471-1478, 2004
Krak NC, Boellaard R, Hoekstra OS, et al: Effects of ROI definition and reconstruction method on quantitative outcome and applicability in a response monitoring trial. Eur J Nucl Med Mol Imaging 32:294-301, 2005
Vansteenkiste JF, De Leyn PR, Deneffe GJ, et al: Clinical prognostic factors in surgically treated stage IIIA-N2 non-small cell lung cancer: Analysis of the literature. Lung Cancer 19:3-13, 1998
Keyes JW Jr: SUV: Standard uptake or silly useless value? J Nucl Med 36:1836-1839, 1995
Krak NC, van der Hoeven JJM, Hoekstra OS, et al: Measuring [(18)F]FDG uptake in breast cancer during chemotherapy: Comparison of analytical methods. Eur J Nucl Med Mol Imaging 30:674-681, 2003
Kroep JR, van Groeningen CJ, Cuesta MA, et al: Positron emission tomography using 2-deoxy-2-[18F]-fluoro-D-glucose for response monitoring in locally advanced gastroesophageal cancer: A comparison of different analytical methods. Mol Imaging Biol 5:337-346, 2003(Corneline J. Hoekstra, Si)