Predictive and prognostic value of FDG-PET in nonsmall-cell lung cancer

A systematic review

Authors

  • Lioe-Fee de Geus-Oei MD,

    Corresponding author
    1. Department of Nuclear Medicine, Radboud University Nijmegen Medical Center, Nijmegen, Netherlands
    • Department of Nuclear Medicine (internal postal code 444), Radboud University Nijmegen Medical Centre, P.O. Box 9101, 6500 HB Nijmegen, Netherlands
    Search for more papers by this author
    • Fax: (011) 31-24-3618942

  • Henricus F.M. van der Heijden MD, PhD,

    1. Department of Pulmonary Diseases, Radboud University Nijmegen Medical Center, Nijmegen, Netherlands
    Search for more papers by this author
  • Frans H.M. Corstens MD, PhD,

    1. Department of Nuclear Medicine, Radboud University Nijmegen Medical Center, Nijmegen, Netherlands
    Search for more papers by this author
  • Wim J.G. Oyen MD, PhD

    1. Department of Nuclear Medicine, Radboud University Nijmegen Medical Center, Nijmegen, Netherlands
    Search for more papers by this author

Abstract

For several years, molecular imaging with 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) has become part of the standard of care in presurgical staging of patients with nonsmall-cell lung cancer (NSCLC), focusing on the detection of malignant lesions at early stages, early detection of recurrence, and metastatic spread. Currently, there is an increasing interest in the role of FDG-PET beyond staging, such as the evaluation of biological characteristics of the tumor and prediction of prognosis in the context of treatment stratification and the early assessment of tumor response to therapy. In this systematic review, the literature on the value of the evolving applications of FDG-PET as a marker for prediction (ie, therapy response monitoring) and prognosis in NSCLC is addressed, divided in sections on the predictive value of FDG-PET in locally advanced and advanced disease, the prognostic value of FDG-PET at diagnosis, after induction treatment, and in recurrent disease. Furthermore, the background and recommendations for the application of FDG-PET for these indications will be discussed. Cancer 2007. © 2007 American Cancer Society.

Nonsmall-cell lung cancer (NSCLC) is the leading cause of cancer-related death in both men and women.1 Surgery with curative intent represents the best chance for cure, but is only an option in patients with stages I, II, and selected cases of stage III. However, only 30% of NSCLC cases present at early stages.2 Even if a complete curative resection can be performed, about 50% of patients will recur without adjuvant chemotherapy treatment.3 The majority of these recurrences occur at distant sites. Hence, each pathological stage consists of a heterogeneous population containing individuals at much higher risk of recurrence and death than others. Therefore, there is a need for noninvasive functional imaging modalities that could play a role in further characterization of NSCLCs.

Another major clinical problem in the treatment of NSCLC is therapy response monitoring, because the morphologic information provided by chest radiographs and computed tomography (CT) cannot reliably distinguish necrotic tumor or fibrotic scar from residual tumor tissue.4 Response evaluation with these morphological imaging methods do not correlate well with pathologic response, nor with changes at the cellular level or with tumor viability.5 Indeed, final treatment outcome will be determined more by the biological aggressiveness of residual tumor than by its volume.5 Examination of biopsy material also has its limitations, because it can be inconclusive because of sampling difficulties, it provides information at a single timepoint only, and is not suitable for sequential assessment.6

A potential noninvasive molecular imaging tool that could be helpful in solving these problems is positron emission tomography (PET). The most frequently used tracer in PET is 18F-fluorodeoxyglucose (FDG). FDG-PET localizes tumors by identifying cells in the body that have increased glucose uptake and metabolism. FDG is transported into cells analogous to glucose and is converted to FDG-6-phosphate. This metabolite is trapped in the cell, as it will not be processed in the glycolytic pathway, and hence will accumulate preferentially in those cells with high glucose uptake, such as tumor cells.7 The introduction of the combined use of FDG-PET and CT has had a major impact on the diagnosis and staging of lung cancer. FDG-PET has been used to evaluate unclassified pulmonary nodules for malignancy.8, 9 Furthermore, it provides noninvasive mediastinal staging and reduces the number of futile thoracotomies and mediastinoscopies.10–12 This imaging modality detects unsuspected extrathoracic metastases in 14% to 17% of patients otherwise considered potentially resectable.13 Recently, FDG-PET has also demonstrated its value in radiation treatment planning and detection of recurrent disease.14–17

Because FDG-PET relies on the detection of metabolic alterations of cancer cells, this examination yields data independently of associated structural characteristics, and therefore allows the detection or monitoring of specific metabolic changes that are not associated with or precede the anatomical changes.18 One of the great advantages of this technique is that it can not only observe but can also quantify FDG uptake to distinguish metabolically highly active from less active tumor tissues and therefore offers an opportunity for noninvasive, in vivo tissue characterization. Hence, the number of clinical applications for FDG-PET in NSCLC continues to increase. This comprehensive review aims to discuss the potential future applications in the management of patients with NSCLC. The literature is reviewed on the value of FDG-PET in identifying tumor response to anticancer therapies at an early phase of treatment and in identifying subsets of patients with poor outcome.

Search Strategy and Selection Criteria of Literature

Data for this review were identified by searches of PubMed and SilverPlatter MEDLINE up to July 2006 using the search terms for identifying clinical FDG-PET studies as reported by Mijnhout et al.19 plus the search terms “lung neoplasm or lung cancer or NSCLC” and “prediction or prognosis or therapy monitoring or response monitoring.” Only articles that make use of the exact definition for the terms “predictive” and “prognostic” were included. These exact definitions are as follows: 1) FDG-PET is considered an in vivo marker for prognosis if it is able to identify patients that are at a different risk of a specific outcome, such as tumor progression or death, irrespective of a specific therapy. 2) FDG-PET is considered a predictive in vivo marker if it is able to define patient populations that (will) benefit from a specific form of treatment.20 Search results were checked for adequacy. Only articles published in English were included. Articles only dealing with staging and detection of recurrences of NSCLC with FDG-PET, with other malignancies than NSCLC, and with radiopharmaceuticals other than FDG were omitted. References from adequate articles were checked for studies not retrieved by the search strategy. Abstracts, (extended) case reports, reports from meetings, editorial comments and letters-to-the-editor were not included.

Background and Technical Aspects of (Semi)Quantitative FDG-PET

For the application of FDG-PET in the detection, staging, and surveillance of primary and recurrent tumors, a qualitative image interpretation has been found sufficient in most cases. In contrast, for the application of FDG-PET in chemotherapy response monitoring and prognostic stratification, quantification of FDG uptake will be required to accurately determine tumor metabolism and changes in tumor metabolism over time. Glucose metabolism can be measured mathematically using the glucose flux constants (as used to determine the metabolic rate of glucose or MRglu) or, clinically more feasible, using the standardized uptake value (SUV), a semiquantitative measure of FDG uptake.21

A variety of techniques and analytical methods are currently used, which complicates the comparison of different studies on this subject. It is still unclear whether more advanced kinetic techniques are superior to more basic methods like calculation of the SUV. For calculation of SUVs, only the amount of injected radioactivity and the body weight of the patient are required.21 Many factors may influence FDG uptake, such as plasma glucose, insulin levels, the interval between FDG injection and image acquisition, and a significant loss of body weight during anticancer therapy could affect the use of serial measurements of the SUV. Correction of SUV by body surface area (SUVBSM) or lean body mass (SUVLBM) reduces this dependency on body weight.22, 23 More advanced kinetic techniques and analytical methods, like the Patlak et al. analysis,24 take into account differences in the whole-body distribution of FDG at the time of scanning. Therefore, MRglu is in principle a more reliable measure than SUV.25 However, kinetic techniques require dynamic data acquisition and the clearance of FDG from the blood has to be determined. The different methods are not only of interest from a methodological point of view, but they also determine how readily PET imaging can be implemented as a routine clinical tool for response monitoring. Multiple follow-up scans encourage a simplified approach to improve patient compliance, an important feature of successful clinical trials. Furthermore, during the dynamic data acquisition only an axial field of view of 15–20 cm can be studied, whereas SUVs can also be calculated from whole-body PET studies. As metastatic lesions in different parts of the body may respond differently to chemotherapy, this represents a principal advantage of SUV.25

SUV quantification depends on acquisition, reconstruction, and region of interest parameters and differs between different scanners and different data analysis software.26 Therefore, in treatment response monitoring it is important to perform sequential scans of the same patient on the same scanner and under identical scanning, image reconstruction, and data analysis conditions. Optimal timing of PET scans at some point during the treatment schedule or soon after the completion of treatment depends on different chemosensitivities of tumors, tumor heterogeneity, and modes of drug action.27 Assessments performed too early might overestimate FDG uptake because glucose metabolism can still be present in cells that have received lethal damage, and because of inflammatory reactions in responding tissues. Assessments that are performed too late can also be less suitable, because of the risk of tumor repopulation.5 Especially after radiotherapy it has been frequently recommended6, 28–30 to carefully select the optimal interval from completion of radiotherapy to FDG-PET imaging. At present, there are no systematic data available on patients with NSCLC to determine the optimal time to perform FDG-PET after radiotherapy. However, the influence of radiation-induced inflammatory reactions does not seem a major issue because all mentioned reports in a postradiotherapy setting did not show major confounding results.4, 31–39

Predictive Value of FDG-PET and Assessment of Histopathologic Response in Locally Advanced Disease

Nine studies indicated a possible role for PET in assessment of response during or after radiotherapy,31 induction chemotherapy,33, 40 or a combination thereof.4, 32, 36, 37, 41, 42 The earliest study, performed in 1996 by Ichiya et al.,31 reported on 20 patients before and after radiation therapy. Patients with a higher FDG uptake on their baseline PET showed a better response to treatment. The recurrence rate was higher in lesions with higher uptake before as well as after treatment. A more prominent decrease in the FDG uptake was noted in patients with a response on CT compared with those with no change on CT. In the pilot study of Vansteenkiste et al.,40 who studied 15 patients before and after induction chemotherapy, a reduction in FDG uptake of at least 50% in the primary tumor or mediastinal clearance proved to be a better predictor of long-term survival compared with the standard WHO criteria43, 44 used for response assessment on CT. Another study that evaluated response monitoring of induction chemotherapy33 showed that FDG-PET identified prognostically different strata in patients considered responsive according to CT. In this prospective multicenter study FDG-PET was performed before and after 1 and 3 cycles of induction chemotherapy. This was the only study that evaluated FDG-PET at an earlier and perhaps clinically more relevant stage of treatment. The residual metabolic rate of glucose after 1 cycle selected patients with different outcomes. The residual metabolic rate of glucose after completion of induction chemotherapy proved to be the best prognostic factor. The other 6 studies4, 32, 36, 37, 41, 42 showed that FDG-PET was also predictive with respect to therapy outcome of combined modality therapy. Metabolic response proved to be associated with pathological response (Table 1).32, 37, 41, 42 The accuracy of a dynamic FDG-PET scan in the prediction of pathological tumor response was 83% to 96%.32, 37, 41 There was a near linear relation between the change in SUV and the percentage of nonviable tumor cells in the resected tumors41 and the percentage decrease in SUV in the primary tumor during induction therapy was significantly larger in patients with <10% than in patients with >10% residual tumor cells.42 The SUV in the mediastinal lymph nodes after completion of chemoradiotherapy predicted the histopathologic lymph node status after radiochemotherapy (ypN0 status) with a sensitivity and specificity of 73% and 89%.42 Cerfolio et al.41 showed that metabolic response correlated better with pathology than the change in size on CT scan. In a study performed by Mac Manus et al.4 there was poor agreement between PET and CT responses, which were identical in only 40% of patients. There were significantly more complete responders on PET than CT, whereas fewer patients were judged to be nonresponders or nonassessable, which implies that chemoradiation may be a more effective therapy than previous CT-based assessments have suggested. On multifactor analysis (including the known prognostic factors of CT response, performance status, weight loss, and stage), only the PET response was significantly associated with the duration of survival. Another study performed by Mac Manus et al.36 showed a significantly longer median survival for patients with complete metabolic response than for patients with noncomplete metabolic response (31 vs 11 months). Noncomplete metabolic response patients had higher rates of local failure and distant metastasis than patients with complete metabolic response. Significantly more patients with a larger percentage decrease in SUV in the primary tumor during induction therapy stayed free from extracerebral recurrence compared with patients with a lesser response (83% vs 43% at 16 months).42

Table 1. Sensitivity and Specificity of FDG-PET for Assessing Histopathologic Response
ReferenceSensitivitySpecifityDefinition of metabolic responseDefinition of histopathologic response
  1. SUV indicates standardized uptake value; PET, positron emission tomography; FDG, 18F-fluorodeoxyglucose; MRglu, metabolic rate of glucose.

3286%81%Classified as metabolic responder if predicted probability >0.5 Post-hoc definition: No complete pathologic response: residual MRglu ≥0.130 μmol/min/gNo pathologic response: positive resection margins or gross residual tumor in the resected specimen
Complete pathologic response: residual MRglu ≤0.050 μmol/min/gPathologic response: microscopic residual disease only or a pathologically complete response
3788%89%Post-hoc definition for metabolic response: SUV ≤4.5 on the PET after therapyNo pathologic response: residual macroscopic disease.
Pathological response: no tumor (complete response) or residual microscopic disease only
4190%100%Post-hoc definition for metabolic response: >80% reduction of the maximum SUVComplete pathologic response: ≤1% of viable tumor cells
4280%80%Post-hoc definition for metabolic response: SUV ≤3.3 after therapyGrade 1, no or only spontaneous tumor regression; Grade 2a, evidence of therapy-induced tumor regression with >10% residual tumor cells; Grade 2b, evidence of therapy-induced tumor regression with <10% residual tumor cells; Grade 3, no evidence of vital tumor
3888%67%Post-hoc definition for metabolic response: SUV ≤3.0 after therapyResidual cancer: resection specimen positive for malignant cells
Pathologic complete response: resection specimen negative for malignant cells
6881%64%Post-hoc definition for metabolic response: SUV ≤2.5 after therapyResidual cancer: resection specimen positive for malignant cells
Pathologic complete response: resection specimen negative for malignant cells
7397%67%Visual analysis: Residual cancer: residual FDG-uptake in the primary tumorResidual cancer: resection specimen positive for malignant cells Pathologic complete response: resection specimen negative for malignant cells
Metabolic complete response: no residual FDG-uptake in the primary tumor

Despite the finding that these 9 studies were very heterogeneous with respect to the applied methods of PET quantification, the primary targets of PET evaluation (primary tumor and/or lymph nodes), and the clinical endpoints (histology, survival), all studies showed that FDG-PET is a significant predictor of therapy outcome and provides results of great prognostic significance. It seems that FDG-PET is able to predict pathological response more accurately and at earlier timepoints than conventional imaging methods.

Predictive Value of FDG-PET in Advanced Disease

Early prediction of tumor response is of particular interest in patients with advanced stages of NSCLC. Tumor progression during first-line chemotherapy occurs in approximately 30% of patients.45 Thus, a significant percentage of patients undergo toxic therapy during several weeks without benefit. As more options for second-line therapies and new-targeted therapies for advanced disease become available, there is a growing need to find a reliable response-monitoring tool. The experience with FDG-PET in this field of NSCLC, however, is still limited. A recent study performed by Lee et al.46 claimed that a sole pretreatment FDG-PET scan is sufficient to predict response and survival for platinum-based combination chemotherapy in chemonaive patients with advanced NSCLC. They found that patients with a high SUV exhibited significantly higher response rates. Other factors, including sex, age, histology, performance status, number of involved organs, regimens used, and disease stage, did not affect response. Regrettably, high SUVs were related with a shorter response duration and shorter time to progression. The authors postulate that higher response rates in patients with hypermetabolic lesions can be explained in terms of the rapid proliferation of tumor cells47, 48 and the effectiveness of the chemotherapy against proliferating cells.49, 50 At the same time, these cells are more aggressive, which is reflected by the shorter response duration and the shorter time to progression. Lee et al.46 mentioned that response rates are often considered as a surrogate endpoint for survival51; however, they are of the opinion that this is not a recommendable surrogate endpoint because there might be a discordance. Weber et al.,25 however, came to other conclusions. The only properly performed study on response assessment in advanced stages (stage IIIB or IV) using serial PET-scans was performed by them. Fifty-seven patients scheduled to undergo platinum-based chemotherapy were studied before and after the first cycle of therapy. A decrease in SUV of 20% or more after 1 cycle of chemotherapy was associated with a longer time to progression (163 days vs 54 days) and longer median overall survival time (252 days vs 151 days). The 1-year survival rate was also significantly higher in metabolic responders compared with nonresponders (44% vs 10%). This study showed that a reduction of metabolic activity already after 1 cycle of chemotherapy is closely correlated with final outcome of therapy. These results suggest that PET imaging may be used to personalize the use of chemotherapy for individual patients and therapeutic regimens can be altered early in the course of therapy to deliver a more favorable outcome and reduce the morbidity and costs associated with prolonged, ineffective chemotherapy. Besides, using metabolic response as an endpoint may shorten the duration of phase 2 studies evaluating new cytotoxic drugs.

Prognostic Value of FDG-PET at Diagnosis

The tumor-node-metastasis (TNM) staging system is considered the most important tool to estimate prognosis and to date is the most important guide in treatment decisions.52 However, the TNM staging system provides an incomplete biologic profile of NSCLC, does not always provide a satisfactory explanation for differences in recurrence and survival, and is thus far from perfect as a prognostic indicator.53 Quantitative measures of biologic aggressiveness, like FDG uptake, could be better indicators for survival and risk of recurrence, and thus for selection of patients for adjuvant treatment.54, 55 For prognostic stratification the SUV can be calculated using a single whole-body FDG-PET that is routinely performed in patients with NSCLC as part of their pretherapeutic staging procedure. A great advantage of measurement of FDG uptake is that this can be done before any treatment has been performed.

The prognostic value of FDG-PET at diagnosis has been evaluated in several studies.53, 56–65 These studies have shown that the pretherapeutic FDG-PET not only improved patient staging, but also provided prognostic information (Table 2). All 11 studies showed that patients with low FDG uptake values in their primary tumor have a significant longer overall- and progression-free survival than patients with high FDG uptake. Several studies56, 57, 61 found that in patients with high FDG uptake prognosis was further reduced if the tumor also exceeded 3 cm in size. Higashi et al.53 and Sasaki et al.62 found that FDG uptake in the primary tumor was a better prognostic variable than pathologic TNM system staging in predicting recurrence of patients with NSCLC. Multivariate Cox analysis, including factors like disease stage, performance status, histology, tumor cell differentiation, tumor size, lymph node involvement, completeness of resection, etc, identified the SUV in the primary tumor as an independent prognostic factor in several studies.57, 60, 61, 63–65 In contrast to FDG uptake in the primary tumor, the prognostic ability of the SUV for the regional lymph nodes remains uncertain. Sasaki et al.62 observed that patients with high SUVs of their regional lymph nodes and low SUVs in their primary tumors did not experience any local or distant recurrence. Therefore, it is at least speculated that the SUVs for the regional lymph nodes do not agree with and are not stronger prognostic factors than the SUVs for the primary tumor.

Table 2. Prognostic Value of FDG-PET at Diagnosis
Yearn*Histology Stage Treatment Survival PReference
  • SUV indicates standardized uptake value; PET, positron emission tomography; FDG, 18F-fluorodeoxyglucose.

  • *

    Number of patients.

1998155Squamous37%I or II45%Unspecified Median survival, mo .004956
 Adeno34%III35%  SUV>1011.4  
 Large-cell7%IV20%  SUV≤1024.6  
 Undetermined22%        
1999125Squamous54%I37%Resection73%2-y survival .00157
 Adeno25%II15%Nonsurgical27%SUV>743%  
 Large-cell21%IIIA30%  SUV≤783%  
   IIIB18%      
199938Squamous32%I19%Resection76%3-y survival .225658
 Adeno50%II13%Nonsurgical24%SUV>8.7240%  
 Large-cell18%IIIA50%  SUV≤8.7270%  
   IIIB5%      
   IV13%      
200077Squamous58%All stages ≤IIIA Not reported Median survival, mo .00159
 Adeno23%    SUV>206  
 Large-cell13%    SUV≤2033  
 Other5%        
200257Squamous14%I81%Resection100%5-y survival .000253
 Adeno60%IA67%  SUV>520%  
 Large-cell2%IB14%  SUV≤590%  
 BAC23%II2%  5-y disease-free survival (stage I) <.0001 
 Adenosquamous2%III17%  SUV>517%  
       SUV≤588%  
           
200273Squamous51%I44%Resection92%2-y survival .001160
 Adeno41%II23%Nonsurgical8%SUV≥756%  
 Large-cell3%IIIA7%  SUV<796%  
 BAC5%IIIB19%      
   IV7%      
2004100Squamous24%All T1-4, N0-2, M0 R0 resection100%2-y survival <.0161
 Adeno67%    SUV>968%  
 Large-cell3%    SUV<996%  
 Adenosquamous2%        
 Carcinoid4%        
2005162Squamous43%I40%Resection57%2-y survival .0262
 Adeno or large-cell46%II16%Radical radiotherapy43%SUV>565%  
 Other10%IIIA20%  SUV≤594%  
   IIIB24%      
2005315Squamous54%IA19%Resection71%Mean survival, y <.00163
 Adeno33%IB26%Nonsurgical29%SUV≥101.6  
 Other13%II18%  SUV<103.2  
   IIIA23%      
   IIIB5%      
   IV9%      
200551Squamous33%I41%Radical radiotherapy100%2-y survival <.00164
 Adeno25%II22%  SUV≥1527%  
 Large-cell20%III37%  SUV<1560%  
 Other22%        
2006137Squamous45%IIIA43%Chemoradiation100%Median survival, mo .0565
 Adeno29%IIIB57%  SUV>129  
 Large-cell12%    SUV<1222  
 Other14%        

The univariate analyses performed to determine a cutoff point for the SUV in the primary tumor to discriminate between a more or less favorable prognosis has ranged widely from 5 to 20. Sasaki et al.62 and Higashi et al.53 reported that for the SUV a group dichotomy with a cutoff value of 5 had the best discriminative value for prognosis. Jeong et al.60 as well as Vansteenkiste et al.57 found a cutoff value of 7, Downey et al.61 a cutoff of 9, Ahuja et al.56 and Cerfolio et al.63 a cutoff of ≈ 10, Eschmann et al.65 of 12, and Dhital et al.59 a cutoff of 20 (Table 2). Vansteenkiste et al.57 and Higashi et al.53 showed that dichotomization with a broad range of SUVs gave significantly discriminative log-rank probability values. Therefore, this implies that the relation between SUV and prognosis could be a gradual one rather than based on a threshold. It seems reasonable to hypothesize that there is no true cutoff point but, rather, a transition zone, within which the prognosis gradually worsens. However, the wide range of SUV values seen in these studies can also be due to the heterogeneity of the analyzed patient cohorts and to variation in the PET-scanners and acquisition protocols used. Institutional-based technical factors can lead to variations in measurement of SUV and might hinder the integrated or comparative interpretation of the results from one center to another.26, 66 For example, in the above-mentioned studies there was considerable variability in the period between tracer administration and scanning, as well as in the serum glucose levels. There was also significant variation in reconstruction and interpretive criteria, such as the lack of correction for partial volume effects in certain studies. Most of the studies chose to use the maximum value of the SUV within the tumor (ie, SUVmax) to avoid inadvertent bias. Therefore, standardization of acquisition, reconstruction, and region of interest (ROI) methods is preferred for SUV quantification in multicenter trials.26 Agreement on methods of scanning, SUV measurement, and the best cutoff values is needed before this technique can be fully exploited in clinical medicine to select patients for adjuvant treatments. In the most ideal situation it would be desirable to study a homogeneous patient cohort in which all patients have the same tumor stage, the same tumor histology, and are treated by the same therapeutic protocol to objectively assess the prognostic information provided by FDG uptake. Nevertheless, despite these variability and concerns, all 11 studies arrived at the same conclusion and strongly confirmed that the degree of tumor glucose use on an FDG-PET scan provide independent prognostic information.

Prognostic Value of FDG-PET After Induction Treatment

The prognostic value of FDG-PET after induction treatment has not been studied as thoroughly as at diagnosis. There is, however, increasing interest in determining the prognostic value of the posttherapy FDG-PET. Several studies (Table 3) have shown promising results, indicating that the posttherapy FDG-PET also has an independent prognostic value and is superior to that of CT.4, 33, 67, 68 The combination of platinum-based chemotherapy and radiotherapy is a commonly recommended standard curative approach in unresectable locally advanced disease.69 Whether or not a pathologic complete response can be achieved is predictive for a more or less favorable outcome.70, 71 Therefore, the pathologic complete response rate is an important endpoint for studies that evaluate new induction treatments. Especially when data of the resection specimen are lacking, accurate noninvasive assessment of pathologic complete response is important. Although pathologic complete response is mostly seen in patients who achieve major tumor shrinkage on CT, it is difficult to differentiate residual tumor from scar tissue in a residual mass on CT.72 Several studies showed that FDG-PET could be very helpful in this condition.38, 73–75 PET response was correlated with pathologic response in 5 studies,38, 40, 68, 74, 75 and 5 studies also reported on the correlation of PET response and survival.4, 33, 40, 67, 68 Multivariate stepwise analysis in the multicenter trial of Hoekstra et al.33 showed that PET identified prognostically different strata in patients considered responsive according to CT.33 Unanimously, all studies on this subject (Table 3) show that the degree of residual tumor FDG uptake after induction therapy is, like tumor FDG uptake at diagnosis, a strong prognostic factor. Patients in whom treatment results in complete resolution of prior FDG uptake have been shown to have a good prognosis as compared with those with residual FDG uptake after induction treatment.

Table 3. Prognostic Value of FDG-PET After Induction Treatment
Yearn*Histology Stage Treatment Survival/Hazard ratio PReference
  • SUV indicates standardized uptake value; PET, positron emission tomography; FDG, 18F-fluorodeoxyglucose; MRglu, metabolic rate of glucose.

  • *

    Number of patients.

2000113Squamous37%I29%Chemotherapy37%3-y survival .00267
 Adeno38%II8%Surgery55%Positive FDG-PET20%  
 Large-cell6%IIIA23%Radiotherapy61%Negative FDG-PET90%  
 BAC6%IIIB23%  Risk ratio positive FDG-PET10.3  
 Other13%IV17%      
200373Squamous62%I18%Radical radiotherapy14%Hazard ratio positive FDG-PET4.2.00044
 Adeno23%II19%Chemo + radical radiotherapy86%    
 Large-cell12%III63%      
 Adenosquamous3%        
200447Squamous51%IIB6%Resection83%Median survival, mo .000668
 Adeno43%IIIA21%Nonsurgical17%SUV≥419  
 Large-cell2%IIIB72%  SUV<456  
 Other4%        
200547Squamous41%All IIIA-N2 Resection53%Hazard ratio  33
 Adeno34%  Nonsurgical47%MRglu>0.13   
 Large-cell23%    MRglu≤0.13   
 Other2%    After 1 cycle3.03.0099 
       After 3 cycles4.10.0009 

Prognostic Value of FDG-PET in Recurrent Disease

The value of follow-up after primary treatment of NSCLC is a matter of debate.76, 77 One may argue that recurrent NSCLC has a poor prognosis, but several studies have shown that retreatment after surgery may increase survival in selected cases.77–79 After potentially curative therapy of NSCLC, masses or symptoms suggestive of recurrence are common but may be difficult to characterize. The application of FDG-PET for the detection of relapsing lung cancer is increasingly being investigated.6, 80–87 In 2 of these studies (Table 4), FDG-PET has provided prognostic information in NSCLC.84, 86 The study of Hellwig et al.86 showed that FDG-PET helps in the selection of patients who will benefit from surgical retreatment. In their study FDG-PET accurately detected recurrent lung cancer (sensitivity, specificity, accuracy: 93%, 89%, 92%) and the SUV in recurrent tumor tissue was an independent prognostic factor. SUV in recurrent tumor was significantly higher than in benign posttherapeutic changes (10.6 ± 5.1 vs 2.1 ± 0.6) and median survival was significantly longer for patients with lower FDG uptake in recurrent tumor. In patients treated surgically, lower FDG uptake predicted longer median survival (SUV <11 to 46 months, SUV ≥11 to 3 months). Hicks et al.84 studied patients with suspected recurrence >6 months after definitive treatment. PET had a sensitivity of 98%. No disease was evident during a minimum follow-up of 12 months in 14 of 15 patients with clinically suspected recurrence but negative PET findings (negative predictive value, 93%). Both the presence (P = .01) and the extent (P < .0001) of recurrence on PET were highly significant prognostic factors. There was also significant prognostic stratification based on the treatment delivered after the PET study, but after adjustment for this treatment PET status remained highly predictive of survival. The authors concluded that PET better assessed the status of disease and PET better stratified prognosis compared with conventional staging.

Table 4. Prognostic Value of FDG-PET in Recurrent Disease
Yearn*Histology Stage Treatment Survival/Hazard ratio PReference
  • SUV indicates standardized uptake value; PET, positron emission tomography; FDG, 18F-fluorodeoxyglucose.

  • *

    Number of patients.

200163Squamous57%Recurrent disease67%Resection48%1-y survival .01284
Adeno30%No relapse33%Radical radiotherapy52%Positive FDG-PET55%  
Large-cell11%    Negative FDG-PET84%  
Other2%    Hazard ratio positive FDG-PET2.95.012 
200662Squamous53%Recurrent disease70%Resection31%Median survival, mo <.0186
Adeno30%Second primary19%Nonsurgical69%SUV≥119  
Large-cell2%No relapse11%  SUV<1118  
BAC8%        
Carcinoid2%        
Undetermined5%        

Conclusions and Recommendations

It has been shown that the degree of FDG uptake is of prognostic value at initial presentation, after induction treatment before resection, and in case of recurrence. At initial presentation, as well as posttreatment, FDG-PET is a better predictor of survival than TNM system staging or CT response. Agreement on methods of scanning, SUV measurement, and the best cutoff values is needed before this technique can be fully exploited in clinical medicine to select patients for adjuvant treatments. In future multicenter trials standardization of acquisition, reconstruction, and ROI methods is preferred for SUV quantification. Further research in this field is of great importance because it may induce a change in the therapeutic concept of patients with NSCLC. FDG-PET could separate patients with a good prognosis from those with a poor one, which may help in deciding on the most appropriate treatment and which may be of particular value in stratifying patients for clinical trials.

FDG-PET is also of value in predicting outcome of induction therapy and it probably also has a predictive value early in the course of first-line therapy in the case of advanced disease. There are indications that FDG-PET can predict response and patient outcome as early as after 1 course of (induction) chemotherapy. However, monitoring tumor response with FDG-PET is still in its infancy. The methods of measurement of FDG uptake currently are diverse and timing with respect to anticancer therapy and used thresholds to define response are variable. Therefore, further study is required to deal with these major issues before it is possible to draw definite conclusions on FDG-PET as a tool for therapy response monitoring. If the results of the reviewed studies can be confirmed, FDG-PET could shorten the track of early clinical trials that assess new antineoplastic agents and could also improve patient management by reducing morbidity, efforts, and costs of ineffective treatment in nonresponders.

Ancillary