Sara Sheikhbahaei 1 Mehdi Taghipour 1 Rubina Ahmad 1 Carole Fakhry 2,3 Ana P. Kiess 4 Christine H. Chung 5 Rathan M. Subramaniam 1,2,5,6

Transcription

1 Nuclear Medicine and Molecular Imaging Original Research Nuclear Medicine and Molecular Imaging Original Research Sara Sheikhbahaei 1 Mehdi Taghipour 1 Rubina Ahmad 1 Carole Fakhry 2,3 Ana P. Kiess 4 Christine H. Chung 5 Rathan M. Subramaniam 1,2,5,6 Sheikhbahaei S, Taghipour M, Ahmad R, et al. Keywords: FDG PET,, head and neck neoplasm, meta-analysis, PET/CT, recurrence, sensitivity and specificity DOI: /AJR Received November 27, 2014; accepted after revision February 19, Russell H. Morgan Department of Radiology and Radiological Sciences, Johns Hopkins University School of Medicine, 601 N Caroline St, JHOC 3235, Baltimore, MD Address correspondence to R. M. Subramaniam (rsubram4@jhmi.edu). 2 Department of Otolaryngology and Head and Neck Surgery, Johns Hopkins University School of Medicine, Baltimore, MD. 3 Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD. 4 Department of Radiation Oncology and Radiation Molecular Sciences, Johns Hopkins University School of Medicine, Baltimore, MD. 5 Department of Medical Oncology, Johns Hopkins University School of Medicine, Baltimore, MD. 6 Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD. AJR 2015; 205: X/15/ American Roentgen Ray Society Diagnostic Accuracy of Follow-Up FDG PET or PET/CT in Patients With Head and Neck Cancer After Definitive Treatment: A Systematic Review and Meta-Analysis OBJECTIVE. The purpose of this study was to assess the diagnostic performance of FDG PET or PET/CT for detection of local, regional, and distant recurrences in the of patients with head and neck cancer who underwent definitive treatment. MATERIALS AND METHODS. A systematic search was performed in MEDLINE and Cochrane Library (updated September 2014) to identify relevant published studies. Studies investigating the accuracy of FDG PET/CT that were performed at least 4 months after therapy were included. Two authors independently screened all retrieved articles, selected studies that met the inclusion criteria, and extracted the data. Histopathologic confirmation or of at least 6 month (or both) was considered as the reference standard. RESULTS. Twenty-three studies constituting a total of 2247 FDG PET/CT examinations met our inclusion criteria. The pooled sensitivity and specificity of PET/CT for detection of recurrence were 0.92 (95% CI, ), and 0.87 (95% CI, ), respectively. The pooled sensitivity and specificity of scans performed 4 12 months after treatment were 0.95 (95% CI, ) and 0.78 (95% CI, ), respectively. Similar estimates for scans performed at or more than 12 months after treatment were 0.92 (95% CI, ) and 0.91 (95% CI, ), respectively. The overall accuracy of FDG PET/CT in detecting recurrence is higher in patients without suspicion of recurrence before the scan compared with the patients with suspected recurrence. CONCLUSION. The high diagnostic performance of FDG PET/CT in detecting recurrence in curatively treated patients with head and neck cancer supports its use in clinical practice for patient. Further studies are needed to evaluate the prognostic utility of PET/CT in the of head and neck cancer. H ead and neck cancer represents the sixth leading cancer worldwide with an annual estimate of 12,000 deaths in the United States [1, 2]. Locoregional and distant recurrences occur in about 25 50% of patients with advanced-stage head and neck cancer, predominantly within the first 3 years after treatment, and are associated with significant morbidity and mortality in these patients [3, 4]. Recently, it has been shown that negative human papillomavirus status and undergoing salvage surgery are associated with significant overall survival advantage in recurrent oropharyngeal squamous cell carcinoma [5]. Hence, early detection of recurrence is of value to guide clinicians in the management of these patients [6]. This underscores the role of posttreatment imaging to identify recurrences at earlier stages [6, 7]. CT and MRI are often inconclusive or misleading in differentiation of tumor recurrence from postsurgical or chemoradiotherapy inflammation and fibrosis [8, 9]. FDG PET can identify recurrences that may be missed on anatomic imaging modalities by providing physiologic or functional information [6, 10, 11]. However, FDG PET findings in posttherapy assessment of head and neck cancer are time and therapy dependent [10, 12]. For minimization of chemoradiotherapy-related inflammatory FDG uptake, FDG PET or PET/CT is recommended to be performed at minimum 12 weeks after treatment completion [10]. Previous meta-analyses [3, 12] also reported lower sensitivity and specificity for FDG PET/CT examinations performed within 12 weeks after treatment compared with those performed more than 12 weeks after treatment. Recent studies have suggested that FDG PET/CT also yields good diagnostic performance in long-term surveillance, imparts added value to clinical assessment, and can detect AJR:205, September

2 recurrence, when performed with or without clinical suspicion during [7, 13 19], in patients with head and neck cancer. The aim of this systematic review and meta-analysis is to establish summary estimates of diagnostic performance in FDG PET/CT examinations performed for recurrence detection in patients with head and neck cancer by documentation of location of recurrence, time after definitive therapy (4 12 vs 12 months), and whether or not recurrence was suspected clinically before the FDG PET/CT examination. Materials and Methods Search Strategy Systematic electronic searches of the MEDLINE and Cochrane libraries were performed to identify relevant published studies evaluating the diagnostic performance of FDG PET/CT in detecting recurrence during posttreatment of patients with head and neck cancer. The Boolean search strategy relied on the following keyword combinations: PET OR Positron emission tomography OR positron emission tomography and computed tomography OR F-FDG OR 18 f fluorodeoxyglucose AND head and neck neoplasm OR squamous cell carcinoma of head and neck OR HNSCC OR oral cavity cancer OR oropharyngeal cancer OR hypopharyngeal cancer OR laryngeal cancer OR nasopharyngeal cancer AND surveillance OR OR locoregional recurrence OR second primary neoplasm OR distant metastasis AND diagnostic accuracy OR diagnostic performance OR sensitivity OR specificity. No beginning date limit was used, and the search was updated on September 25, Criteria for Study Consideration Patients We included all treated patients with head and neck cancer who were investigated with an index test and a reference standard, irrespective of the staging of the initial disease. Patients could be asymptomatic (without clinical suspicion of recurrence) or have suspicion of recurrence (with clinical suspicion or equivocal or indeterminate findings at imaging) before the FDG PET/CT study was performed. However, patients with any histologically proven recurrence before the index study were excluded. Index test and reference standard FDG PET or FDG PET/CT studies performed more than 4 months after completion of treatment were used as index tests. Histopathologic confirmation or clinical or imaging (or combination of both) of at least 6 months after FDG PET/CT was considered as the reference standard. Target conditions Detection of local recurrence, regional recurrence, distant metastasis or second primary tumor, or combination thereof were the target conditions in this study. Local recurrence was defined as tumor recurrence at or near the primary site, whereas regional recurrence was defined as tumor infiltration of cervical lymph nodes. 751 Records identified through database searches 797 Records screened from the title and abstract 96 Full-text articles retrieved for assessing eligibility 23 Studies included in qualitative synthesis 23 Studies included in quantitative synthesis (meta-analysis) Selection of Studies, Data Extraction, and Management All records identified from the electronic search were independently reviewed by two postdoctoral research fellows in radiology. Non-English publications, review articles, editorials, and case reports, as well as irrelevant citations, were excluded during the initial assessment. Then, full texts of the remaining articles were retrieved. All potentially eligible articles were independently checked for predefined inclusion and exclusion criteria by two authors, and disagreement was resolved by consensus. Studies investigating other radiotracers; evaluating diagnostic performance of an index test in initial staging, pretreatment setting, or therapy response assessment (< 4 months after treatment); and containing inadequate data to construct 2 2 tables were excluded. Only articles with no overlapping study period were included when there was more than one published article from the same institution. If studies performed several FDG PET/CT examinations, only those performed at least 4 months after completion of treatment were included [7, 11, 14, 20] (Fig. 1). Two researchers independently extracted the following data and discrepancies were resolved in consensus; bibliographic details, patients sampling and characteristics, number of patients or scans, index test, timing of scans performed, target conditions, and reference standard. The raw data were extracted from each study to construct 2 2 contingency tables (cross-relating index test results of the reference standards) to assess the outcome. The contingency tables included the number of true-positive, false-positive, true-negative, and false-negative patients or scans. Assessment of Methodologic Quality The methodologic qualities of the eligible studies were assessed using a modified version of the QUADAS tool (QUADAS-2) [21] as recommended by Cochrane Collaborations [22]. The tool consists of four domains covering study selection, index test, reference standard, and flow and timing. Each domain appraises the risk of bias and applicability of the study through a series of questions (Appendix 1). Two radiologists independently assessed study quality. Disagreements were mediated by further discussion and consensus. Statistical Analysis and Data Synthesis The sensitivity and specificity, along with 95% CIs, were calculated in each study. For display of the variations in the results of the individual studies, coupled forest plots and summary ROC (SROC) curves were created using Revman (version 5.3, ReviewManager). For quantification of the degree of heterogeneity in studies, an I-squared index (I 2 ) and its p value were measured [23]. I 2 lies from 0 to 100, and the respective values around 0, 25, 50, and 75, respectively, indicate no, low, moderate, and high heterogeneity among studies [23]. SROC space is defined by sensitivity (y-axis) and specificity (x-axis), and each data point represents 46 Additional records identified through Cochrane Library Fig. 1 Flowchart shows application of study selection criteria. 701 Records excluded: 77 Non English-language publications 146 Review articles, editorials, case reports 249 Reports assessing other neoplasm (esophageal, thyroid, etc.) 231 With no direct link with the main subject 73 Records excluded: Reports assessing FDG PET/CT performance in pretreatment setting (initial staging of patients with head and neck cancer) or in therapy response assessment (examinations performed < 4 months after treatment) of head and neck cancer Insufficient data to construct 2 2 tables 630 AJR:205, September 2015

3 TABLE 1: Summary of Characteristics of the Included Studies Study Authors (Year/Country) Study Design Patient Characteristics Male-to-Female Ratio Age a (y) No. of Scans in Our Analysis Kao et al. [25] (1998/Taiwan) R NPC patients 24: Stokkel et al. [26] (1999/ P Laryngeal and hypopharyngeal cancer patients 43: (4 160) Netherlands) with clinical suspicion of recurrence Li et al. [27] (2001/United States) HNSCC patients with clinical suspicion of (4 36) recurrence Tsai et al. [28] (2002/Taiwan) NPC patients with indeterminate MRI findings 4 mo after treatment 22: > 4 Wong et al. [29] (2002/ United States) R HNSCC patients; imaging performed for routine or in patients with clinical suspicion Time b (mo) 97: c 6.9 Yen et al. [30] (2003/Taiwan) R NPC patients 53: (4 70) Goshen et al. [31] (2005/Israel) R Oral and oropharyngeal SCC patients with clinical 15: (3 6) suspicion of recurrence and inconclusive findings on radiographs Kubota et al. [20] (2004/Japan) R HNC patients with clinical suspicion of recurrence 31: d > 4 Chan et al. [32] (2006/Taiwan) P NPC patients with clinical and imaging suspicion of 22: (6 108) local recurrence Fakhry et al. [33] (2007/France) R HNSCC patients with clinical suspicion of 28: (6 95) recurrence and equivocal findings on CT or MRI Lee et al. [7] (2007/Korea) R HNSCC patients; imaging performed for routine 136: c,d 6 12,12 24, > 24 or in patients with clinical suspicion Salaun et al. [18] (2007/France) R HNSCC patients with negative findings on 23: (7.3 34) CT or MRI Abgral et al. [34] (2009/France) P HNSCC patients with negative findings on 78: (7.2 16) CT or MRI Krabbe et al. [11] (2009/ P Oral cavity or oropharyngeal SCC (stage III or IV) 32: c,d 6, 9, 12 Netherlands) Ng et al. [24] (2010/Taiwan) P NPC patients with high risk or suspicion of 136: (3 56) recurrence Wierzbicka et al. [35] (2011/ R HNSCC patients without any clinical suspicion of 64: (5 22) Poland) recurrence Zundel et al. [36] (2011/United States) R HNC patients 36: (4 6) Ghanooni et al. [37] (2011/ Belgium) Ho et al. [17] (2013/United States) P R Oral cavity, oropharyngeal, hypopharyngeal, and laryngeal SCC patients in all stages HNSCC patients with negative findings at previous imaging (FDG PET/CT or MRI) 23: , : c 12, 24 Kim et al. [14] (2013/Korea) R HNSCC patients 113: d 13 (9 20) Nakamura et al. [16] (2013/ R HNC patients with or without suspicion of 213: , 6 12, 12 24, > 24 Japan) recurrence on clinical and imaging Paidpally et al. [15] (2013/ R HNSCC patients 98: (4 24) United States) Robin et al. [38] (2015/France) R HNSCC patients without any suspicion of recurrence on CT or MRI 97: Note R = retrospective, P = prospective, NPC = nasopharynx carcinoma, SCC = squamous cell carcinoma, HNSCC = head and neck SCC, HNC = head and neck cancer. = not reported. a Reported as average (median or mean). b Interval, reported as median (range) or exact time duration, between completion of treatment and FDG PET/CT imaging. c For some patients, more than one scan per patient was acquired in these studies. d Only patients with PET/CT examinations performed > 4 months after treatment were included in these studies. one particular study. The pooled sensitivity and specificity, along with the 95% CI, were estimated by CMA software (version 2.0, Comprehensive Meta-analysis) using random-effects model. Studies with extractable reports on the site of recurrence (local recurrence, regional recurrence, and distant metastasis or second primary tumor), timing of posttreatment scan (4 12 vs 12 months), and patient setting (with or without clinical suspicion of recurrence) were included in sensitivity analyses. The diagnostic performance of FDG PET/CT in each group was reported. AJR:205, September

4 Results Search Results The process of study selection is diagrammed in Figure 1. The MEDLINE and Cochrane library searches identified 797 articles. Of these, 701 articles were excluded on the basis of the review of the titles and abstracts. We retrieved the full text of 96 relevant articles and subsequently narrowed this down to 23 studies, constituting a total of 2247 FDG PET/CT examinations, that met our inclusion criteria [7, 11, 14 18, 20, 24 38]. Patient Selection Index Test Reference Standard Flow and Timing Risk of Bias (%) Applicability Concerns (%) High risk Unclear risk Low risk Fig. 2 Graph details risk of bias and applicability concerns. Review authors judgments about each domain are presented as percentages across included studies (number of studies within bars). Study Characteristics and the Methodologic Quality Assessment The study design, patient characteristics, index test and interpretation, target condition, and reference standard(s) of each included study are summarized in Tables 1 and 2. The average (mean or median) age of patients in each study ranged from 45 to 63 years. The male-to-female ratio in the studies ranged from 2.0 to 8.6. All studies enrolled previously treated patients with head and neck cancer without histologic evidence of recurrence before scans were acquired. However, the clinical context of patients was different across the included studies. In nine studies, FDG PET/CT examination was performed for patients with clinical (or imaging) suspicion of recurrence before scan acquisition. In five studies, only patients who had negative findings at (physical examination or imaging with CT or MRI) for detection of recurrence were included. In the remaining studies, FDG PET/CT examinations were performed either as a routine or in patients for whom there was clinical suspicion of recurrence. Twelve studies used FDG PET, and 11 used FDG PET/CT as an index test. All except three studies used attenuationcorrected images for PET interpretation; two of these did not report the exact method. In all Kao et al ( ) 0.96 ( ) Stokkel et al ( ) 0.77 ( ) Li et al ( ) 0.86 ( ) Tsai et al ( ) 0.93 ( ) Wong et al ( ) 0.72 ( ) Yen et al ( ) 0.93 ( ) Kubota et al ( ) 0.77 ( ) Goshen et al ( ) 0.67 ( ) Chan et al ( ) 0.83 ( ) Fakhry et al ( ) 0.57 ( ) Lee at al ( ) 0.94 ( ) Salaun et al ( ) 0.95 ( ) Abgral et al ( ) 0.85 ( ) Krabbe et al ( ) 0.72 ( ) Ng et al ( ) 0.90 ( ) Wierzbicka et al ( ) 0.83 ( ) Zundel et al ( ) 0.65 ( ) Ghanooni et al ( ) 0.84 ( ) Ho et al ( ) 0.97 ( ) Kim et al ( ) 0.95 ( ) Nakamura et al ( ) 0.96 ( ) Paidpally et al ( ) 0.91 ( ) Robin et al ( ) 0.87 ( ) Pooled estimate 0.92 ( ) 0.87 ( ) I 2 = 0, p = 0.94 I 2 = 80.14, p < A studies except for the two that did not report exact method [35, 36], visual interpretation was used to determine positivity with regard to recurrence. The reference standard criteria were clearly described across all studies and met our definition (Table 2). Figure 2 graphically presents the risk of bias and applicability concerns in each domain across the included studies according to QUADAS-2 questions. Table 3 shows the number of eligible studies and scans in each subgroup analysis. Findings In the scan level analysis, the pooled sensitivity and specificity of posttreatment FDG Sensitivity Specificity B Fig. 3 Sensitivity, specificity, and accuracy of FDG PET/CT in detection of recurrence in patients with head and neck cancer. A, Forest plot shows sensitivity and specificity across individual studies. In Lee et al. s study, values for locoregional recurrences were incorporated in their analysis. TP = true-positive, FP = false-positive, FN = false-negative, TN = true-negative. I-squared index (I 2 ) represents quantification of heterogeneity from 0 to 100 (low to high, respectively). B, Summary ROC curve analysis outlines diagnostic performance of FDG PET/CT in all studies. Sizes of points indicate sample size in each study. 632 AJR:205, September 2015

5 TABLE 2: Index Test, Target Condition, and Reference Standard of Included Studies Study Authors Index Test a Attenuation Correction Index Test Interpretation FOV Kao et al. [25] FDG PET Yes Visual and quantitative interpretation by 2 experienced nuclear medicine physicians (blind) Stokkel et al. [26] FDG PET No Visual interpretation by 2 experienced observers Li et al. [27] FDG PET Yes Visual and quantitative interpretation by 2 physicians Tsai et al. [28] FDG PET Yes Visual interpretation by 3 experienced nuclear medicine physicians (semiblind) Wong et al. [29] FDG PET Yes Visual and semiquantitative interpretation by 1 of the 5 physicians (not blind) Yen et al. [30] FDG PET Yes Visual interpretation by 2 of 3 experienced nuclear medicine physicians (blind) Goshen et al. [31] FDG PET Yes Visual interpretation by 2 physicians Separate Acquisition b Target Condition(s) Reference Standard(s) Whole body Recurrence Histopathology Middle auditory canal to upper part of chest Inferior orbit to upper femur Recurrence Endoscopic biopsy, 6 mo No Recurrence, residual tumor Histopathology, 6 mo Whole body Recurrence Histopathology, 6 mo Orbitomeatal line to upper abdomen Kubota et al. [20] FDG PET Yes Visual interpretation Vertex to diaphragm Chan et al. [32] FDG PET Yes Visual and semiquantitative interpretation by 3 experienced physicians (semiblind) Fakhry et al. [33] FDG PET Yes Visual interpretation by 2 Lee et al. [7] FDG PET Yes Visual interpretation by 1 physician (blind) Salaun et al. [18] FDG PET Yes Visual interpretation by 2 Abgral et al. [34] FDG PET/CT Yes Visual interpretation by 2 Krabbe et al. [11] FDG PET/CT Yes Visual interpretation by 2 or 3 experienced Ng et al. [24] FDG PET/CT Yes Visual interpretation by 2 physicians and 1 radiologist (semiblind) Whole body Recurrence, residual tumor Whole body 6 mo relapse-free clinical Histopathology, 6 mo Histopathology, CT or MRI, 6 mo clinical Recurrence Endoscopic biopsy, disease-free survival 12 mo Head to upper thigh Local recurrence Histopathology, 6 mo clinical and imaging Whole body Yes Recurrence Histopathology, 8 mo Skull to femur head Locoregional, SP Skull base to subinguinal region Whole body Skull base to mid thigh Head to proximal thigh, SP Histopathology, 6 mo Histopathology, 6 mo Histopathology for locoregional findings, 6 mo of imaging Histopathology, 18 mo clinical and imaging after treatment Histopathology, 12 mo (Table continues on next page) AJR:205, September

6 TABLE 2: Index Test, Target Condition, and Reference Standard of Included Studies (continued) Study Authors Index Test a Attenuation Correction Index Test Interpretation FOV Wierzbicka et al. [35] FDG PET/CT Yes SUV > 3 interpreted as positive by 1 nuclear medicine physician and 1 radiologist Zundel et al. [36] FDG PET/CT Quantitative interpretation by 1 physician (not blind) and 1 radiologist (blind) Ghanooni et al. [37] FDG PET/CT Yes Visual and semiquantitative interpretation by 2 Ho et al. [17] FDG PET/CT Visual and semiquantitative interpretation; equivocal scans considered as negative Kim et al. [14] FDG PET/CT Yes Visual and semiquantitative interpretation by 1 experienced nuclear medicine physician Nakamura et al. [16] FDG PET/CT Yes Visual interpretation by 2 radiologists (not blind) Paidpally et al. [15] FDG PET/CT Yes Visual interpretation by 3 physicians Robin et al. [38] FDG PET/CT Yes Visual interpretation by 2 PET/CT were 0.92 (95% CI, ) and 0.87 (95% CI, ), respectively. Study estimates of sensitivity and specificity were shown using a forest plot (Fig. 3A). The SROC curve summarizing the accuracy of FDG PET/CT examinations across these 23 studies is depicted in Figure 3B. Patient-level analysis was also performed after excluding studies with more than one examination per patient [7, 11, 17, 29]. This analysis yielded fairly similar results, with pooled sensitivity and specificity of 0.91 (95% CI, ) and 0.87 (95% CI, ), respectively. In a subgroup analysis of recurrence sites, the pooled sensitivity of FDG PET/CT for detection of local recurrence, regional recurrence, and distant metastasis or second primary tumor were 0.91 (95% CI, ), 0.88 (95% CI, ), and 0.93 (95% CI, ), respectively. The pooled specificity of FDG PET/CT for detection of local recurrence, regional recurrence, and distant metastasis or second primary tumor were 0.89 (95% CI, ), 0.95 (95% CI, ), and 0.97 (95% CI, ), respectively. The sensitivity distribution appears to be reasonably homogeneous across the studies (I 2 = 0), whereas the specificity values displayed high, moderate, and low heterogeneity for local, regional, and distant recurrences, respectively (Fig. 4). Further, studies were subgrouped by the time interval between completion of treatment and posttherapy FDG PET/CT examination (4 12 vs 12 months). The pooled sensitivities were 0.95 (95% CI, ) and 0.92 (95% CI, ) for examinations performed 4 12 months and at or more than 12 months after treatment (I 2 = 0 for both). However, examinations performed more than 12 months after treatment were more specific compared with those performed 4 12 months after treatment (0.91 [95% CI, ] vs 0.78 [95% CI, ]). The sensitivity and specificity of the studies assessing diagnostic performance of FDG PET/CT 4 12 months and at or more than 12 months after treatment Separate Acquisition b Target Condition(s) Reference Standard(s) Head to knee Regional recurrence Histopathology Skull base to mid thigh Skull base to mid thigh Skull to proximal thigh Yes, SP, SP Head to upper thigh, SP Histopathology, 29 mo of Histopathology, 18 mo Histopathology, subsequent imaging, 6 mo Histopathology, 18 mo clinical and imaging after treatment Histopathology, 6 mo clinical and imaging Chin to mid thigh Yes Recurrence, DM, SP Histopathology, 3 mo Whole body Note DM = distant metastasis, SP = second primary tumor(s), = not reported, SUV = standardized uptake value. a None of the included studies reported the use of high-resolution contrast-enhanced PET/CT. b Whether separate head and neck scan was acquired after whole-body scan. Histopathology, 6 mo are shown in a forest plot and an SROC curve analysis (Figs. 5A and 5B, respectively). According to the SROC curve analysis (Fig. 5B), there was greater variability in the estimates of specificity than sensitivity. Moreover, it is clear that FDG PET/CT examinations performed at or more than 12 months after treatment (red curve in Fig. 5B) dominate across all specificity values, because the curves do not cross each other on SROC curve analysis. We then explored the accuracy of PET/CT on the basis of clinical suspicion before imaging. Figure 6A shows a forest plot of the subgroup analysis according to the clinical context of patients before imaging. Overall, the pooled sensitivities of PET/CT were 0.94 ( ) and 0.89 ( ) in the group with suspected recurrence and the asymptomatic group, respectively. The pooled specificity was higher in the asymptomatic group compared with patients with clinical (or imaging) suspicion of recurrence (0.92 vs 0.78, respectively). There was no to low heterogeneity in the distribution of sensi- 634 AJR:205, September 2015

7 tivity (I 2 < 50) and moderate (I 2 = 50 75) heterogeneity in the distribution of specificity among studies in each group (Fig. 6A). Figure 6B shows that the overall performance of FDG PET/CT is higher in the asymptomatic group compared with the group with suspected recurrence. The SROC curve analysis of FDG PET/CT in patients without clinical (or imaging) suspicion of recurrence before the examination (Fig. 6B) yielded higher diagnostic accuracy of FDG PET/CT in this group compared with those with clinical (or imaging) suspicion of recurrence [39]. Further, sensitivity analyses for the diagnostic performance of FDG PET versus FDG PET/CT to determine the effect of imaging modality on the sensitivity and specificity were performed. The pooled sensitivities were 0.95 (95% CI, ) for FDG PET and 0.91 (95% CI, ) for FDG PET/CT (I 2 = 0 for both). The pooled specificities were 0.85 (95% CI, ) and 0.88 (95% CI, ) for FDG PET and FDG PET/CT, respectively. Imaging modality did not affect the estimates of the sensitivity and specificity. Discussion This systematic review and meta-analysis provides pooled estimates of sensitivity of 0.92 (95% CI, ) and specificity of 0.87 (95% CI, ) for FDG PET/CT in detection of recurrence in patients with head and neck cancer. The SROC curve analysis yielded a good trade-off between sensitivity and specificity (Fig. 3B). Our results for the diagnostic values of FDG PET/CT for detection of local and regional recurrence compared favorably with the previous meta-analysis by Gupta et al. [12], who reported sensitivity and specificity for the primary site of and 0.869, respectively, and for the neck node of and 0.943, respectively, in examinations performed more than 12 weeks after treatment. Another meta-analysis, by Gao et al. [13], including 10 studies constituting 797 PET/CT examinations, reported sensitivity of 0.92 and specificity of 0.95 for screening to detect distant metastasis before salvage treatment in patients with suspected recurrent head and neck cancer, which results are fairly close to ours. Overall, FDG PET/CT was found to be more sensitive for detection of local recurrence and distant metastasis than for detection of regional recurrence. The effect that timing of the imaging examination after therapy had on the diagnostic performance of FDG PET/CT was evaluated by categorizing posttreatment imaging examinations between those performed 4 12 months and at or more than 12 months after treatment; we excluded FDG PET/CT examinations performed within the first 4 months after completion of treatment, because most of these scans were performed to evaluate the treatment response and residual tumor rather than to detect recurrence. In posttherapeutic assessment, the diagnostic accuracy of functional imaging mostly depends on the time interval between the end of the treatment and the imaging examination [10, 12]. This impact has been determined for early FDG PET/CT before 4 months in patients with head and neck cancer [3, 12]. However, there is no meta-analysis that focuses specifically on the performance of long-term imaging examinations in these patients. In both previous meta-analyses by Isles et al. [3] and Gupta et al. [12], the sensitivity of the examinations performed before weeks after completion of treatment was significantly lower than that of examinations performed more than 12 weeks after treatment. Our meta-analysis suggests an improvement in specificity of the FDG PET/CT for detection of recurrences (0.91 vs 0.78), without considerable loss of sensitivity (0.95 vs 0.92), in the examinations performed Local Recurrence Wong et al ( ) 0.79 ( ) Chan et al ( ) 0.83 ( ) Salaun et al ( ) 0.96 ( ) Lee et al ( ) 0.94 ( ) Krabbe et al ( ) 0.84 ( ) Abgral et al ( ) 0.92 ( ) Ng et al ( ) 0.96 ( ) Zundel et al ( ) 0.65 ( ) Kim et al ( ) 0.95 ( ) Robin et al ( ) 0.92 ( ) Pooled estimate 0.91 ( ) 0.89 ( ) I 2 = 0, p = 0.75 I 2 = 81.66, p < Regional Recurrence Wong et al ( ) 0.95 ( ) Salaun et al ( ) 1.00 ( ) Krabbe et al ( ) 0.95 ( ) Abgral et al ( ) 0.98 ( ) Ng et al ( ) 0.96 ( ) Wierzbicka et al ( ) 0.83 ( ) Kim et al ( ) 0.95 ( ) Robin et al ( ) 0.97 ( ) Pooled estimate 0.88 ( ) 0.95 ( ) I 2 = 0, p = 0.65 I 2 = 54.91, p = 0.03 Distant Metastasis or Second Primary Tumor (or Both) Wong et al ( ) 0.96 ( ) Lee et al ( ) 0.98 ( ) Salaun et al ( ) 1.00 ( ) Krabbe et al ( ) 0.92 ( ) Abgral et al ( ) 0.96 ( ) Ng et al ( ) 0.97 ( ) Zundel et al ( ) 0.98 ( ) Kim et al ( ) 0.99 ( ) Robin et al ( ) 0.98 ( ) Pooled estimate 0.93 ( ) 0.97 ( ) I 2 = 0, p = 0.98 I 2 = 27.06, p = 0.20 Fig. 4 Forest plot shows site-specific estimates of sensitivity and specificity of FDG PET/CT across individual studies. TP = true-positive, FP = false-positive, FN = false-negative, TN = true-negative. I-squared index (I 2 ) represents quantification of heterogeneity from 0 to 100 (low to high, respectively). AJR:205, September

8 FDG PET/CT 4 12 mo After Treatment Kao et al ( ) 0.96 ( ) Stokkel et al ( ) 0.60 ( ) Wong et al ( ) 0.72 ( ) Kubota et al ( ) 0.80 ( ) Goshen et al ( ) 0.67 ( ) Lee at al ( ) 0.91 ( ) Krabbe et al ( ) 0.70 ( ) Zundel et al ( ) 0.65 ( ) Ghanooni et al ( ) 0.81 ( ) Nakamura et al ( ) 0.92 ( ) Robin et al ( ) 0.87 ( ) Pooled estimate 0.95 ( ) 0.78 ( ) I 2 = 0, p = 0.99 I 2 = 62.24, p = FDG PET/CT 12 mo After Treatment Stokkel et al ( ) 0.79 ( ) Kubota et al ( ) 0.67 ( ) Lee at al ( ) 0.96 ( ) Krabbe et al ( ) 0.76 ( ) Ghanooni et al ( ) 0.86 ( ) Nakamura et al ( ) 0.98 ( ) Ho et al ( ) 0.97 ( ) Pooled estimate 0.92 ( ) 0.91 ( ) I 2 = 0, p = 0.97 I 2 = 78.4, p < at or more than 12 months compared with 4 12 months after treatment, respectively. The lower specificity in the 4 12-month period may be because of persistent radiationrelated inflammation in some patients, which lasts longer than 12 weeks. It is also possible that there could be a group of patients who are slow responders that is, who have persistent FDG uptake more than 4 months after treatment but have negative results at biopsy. By contrast, in our analysis, six of seven studies [7, 11, 17, 20, 26, 37] with examinations at or more than 12 months after treatment reported sensitivity of 100% for FDG PET/CT in detection of recurrence. The lower pooled result for sensitivity of the examinations performed at or more than 12 months after treatment could be explained by the smaller proportion of scans at or more than 12 months after treatment in each included study. In the current study, curatively treated patients with head and neck cancer were further subgrouped on the basis of the presence of any clinical or imaging suspicion for recurrence before imaging, to quantify the diagnostic performance of FDG PET/CT as a routine surveillance tool (i.e., distinct from its use in the context of preexisting clinical suspicion). The analyses displayed low to moderate heterogeneity between studies in each subgroup. Our results indicated that FDG PET/CT in asymptomatic patients with head and neck cancer with negative findings on with other modalities (CT or MRI) yielded significantly better specificity than when performed in those with suspicion of recurrence. This may be because FDG uptake occurred (and subsequently yielded detection of true-positive malignancy) in sites where findings were equivocal but considered negative on CT or MRI, consequently yielding improved specificity on FDG PET/CT examinations. In contrast, in the group with equivocal but presumed negative findings on CT or MRI, FDG PET/CT examinations yielded higher sensitivity, which may be useful for ruling out disease when negative findings occur in these patients. However, this result should be interpreted with caution, because it is vulnerable to the risk of selection bias. In a previous meta-analysis, Pasamontes Pingarrón et al. [40], reported pooled sensitivity of 0.94 (95% CI, ) and specificity Sensitivity Specificity FDG PET/CT 4 12 mo After Treatment FDG PET/CT 12 mo After Treatment Fig. 5 Sensitivity, specificity, and accuracy of FDG PET/CT according to time interval after completion of treatment. A, Forest plot shows sensitivity and specificity across individual studies. TP = true-positive, FP = false-positive, FN = false-negative, TN = true-negative. I-squared index (I 2 ) represents quantification of heterogeneity from 0 to 100 (low to high, respectively). B, Summary ROC curve analysis outlines effect of timing on FDG PET/CT performance. Sizes of points indicate sample size in each study. A of 0.80 (95% CI, ) for FDG PET in patients with suspected recurrence of head and neck cancer. The respective pooled estimates for sensitivity and specificity in our study are similar to those in Pasamontes Pingarrón et al. s study, although they did not restrict the included studies in their meta-analysis according to the timing of imaging examination relative to treatment. The high sensitivity of FDG PET/CT for the detection of recurrence in patients with clinical (or imaging) suspicion could help in careful selection of eligible patients for tissue diagnosis before therapy. The integration of the separate CT and FDG PET units into a single scanning device has been suggested to have several major interpretive and clinical advantages that could enhance patient management. Greater confidence in interpretation, improved localization of lesions, improved staging of the sentinel lesion, and improved targeting of biopsy and treatment are some of these advantages. In the present meta-analysis, similar to that by Gupta et al. [12], no evidence of considerable difference (according to the 95% CIs of the estimates) was found in the pooled estimates B 636 AJR:205, September 2015

9 Patients With Clinical (or Imaging) Findings Suspicious of Recurrence Stokkel et al ( ) 0.77 ( ) Li et al ( ) 0.86 ( ) Tsai et al ( ) 0.93 ( ) Wong et al ( ) 0.72 ( ) Kabuta et al ( ) 0.89 ( ) Goshen et al ( ) 0.67 ( ) Chan et al ( ) 0.83 ( ) Fakhry et al ( ) 0.57 ( ) Ng et al ( ) 0.90 ( ) Nakamura et al ( ) 0.65 ( ) Paidpally et al ( ) 0.85 ( ) Pooled estimate 0.94 ( ) 0.78 ( ) I 2 = 0, p = 0.82 I 2 = 61.10, p = Patients Without Clinical (or Imaging) Findings Suspicious of Recurrence Salaun et al ( ) 0.95 ( ) Abgral et al ( ) 0.85 ( ) Wierzbicka et al ( ) 0.83 ( ) Paidpally et al ( ) 0.91 ( ) Ho et al ( ) 0.97 ( ) Nakamura et al ( ) 0.98 ( ) Robin et al ( ) 0.87 ( ) Pooled estimate 0.89 ( ) 0.92 ( ) I 2 = 33.10, p = 0.17 I 2 = 74.19, p = of sensitivity and specificity among these two modalities in detection of recurrence of head and neck cancer. However, recent studies have suggested that the complementary use of contrast-enhanced PET/CT and a separate head and neck acquisition could improve the accuracy of FDG PET, particularly in the detection of locoregional recurrences [41]. Further, whether surveillance FDG PET/CT examination improves patients outcomes or just introduces lead-time bias was not evaluated in this meta-analysis of diagnostic accuracy of FDG PET and PET/CT examinations. A growing body of evidence suggests that pretreatment and posttreatment PET/CT volumetric parameters could predict the risk of death and identify patients requiring more aggressive treatment in head and neck cancer [42]. However, a systematic review discussing the utility of PET/CT in predicting adverse events in patients with head and neck cancer has not been performed to date. Our study has some limitations. First, the studies were limited to published English-lan A Sensitivity Specificity Patients with clinical (or imaging) findings suspicious of recurrence Patients without clinical (or imaging) findings suspicious of recurrence Fig. 6 Sensitivity, specificity, and accuracy of FDG PET/CT according to patients clinical setting at time of examination. A, Forest plot shows sensitivity and specificity across individual studies. In Nakamura et al. s study, examinations performed < 4 months after treatment were also included. TP = true-positive, FP = false-positive, FN = false-negative, TN = true-negative. I-squared index (I 2 ) represents quantification of heterogeneity from 0 to 100 (low to high, respectively). B, Summary ROC curve analysis outlines effect of patients clinical setting on FDG PET/CT performance. Sizes of points indicate sample size in each study. TABLE 3: Numbers of Eligible Studies and Imaging Examinations in Each Subgroup Analysis Test No. of Studies No. of Examinations FDG PET/CT for recurrence detection Scan level Patient level Site-specific recurrence Local recurrence Regional recurrence Distant metastasis or second primary tumor Timing of scan after treatment 4 12 months months Clinical (or imaging) findings suspicious of recurrence No Yes Index test FDG PET FDG PET/CT B AJR:205, September

10 guage studies. Exclusion of gray literature, editorials, and articles published in conference proceedings may have introduced publication bias. Second, most of the published studies on this subject were derived from a small group of authors. We restricted the impact of this by excluding studies with overlapping groups of patients from the same institution because that might have influenced the external validity and generalization of our results [43]. In addition, the detailed locations of distant metastasis for truepositive, false-negative, and false-positive PET results were not consistently reported across the included studies. Thus, we were not able to report the incidence and location of distant metastasis and false-positive sites in this study. Conclusion FDG PET/CT yielded high sensitivity and specificity for the detection of recurrence after curative treatment in patients with head and neck cancer. The high diagnostic accuracy of FDG PET/CT in detection of recurrence supports its use in clinical practice for long-term. However, the translation of test performance characteristics into improved patient outcomes has not been addressed in this meta-analysis. Further studies are needed to evaluate the prognostic utility of FDG PET/CT in the surveillance of head and neck cancer. References 1. Siegel R, Ma J, Zou Z, Jemal A. Cancer statistics, CA Cancer J Clin 2014; 64: Kamangar F, Dores GM, Anderson WF. Patterns of cancer incidence, mortality, and prevalence across five continents: defining priorities to reduce cancer disparities in different geographic regions of the world. J Clin Oncol 2006; 24: Isles MG, McConkey C, Mehanna HM. A systematic review and meta-analysis of the role of positron emission tomography in the follow up of head and neck squamous cell carcinoma following radiotherapy or chemoradiotherapy. Clin Otolaryngol 2008; 33: Beswick DM, Gooding WE, Johnson JT, Branstetter BF 4th. Temporal patterns of head and neck squamous cell carcinoma recurrence with positron-emission tomography/computed tomography monitoring. Laryngoscope 2012; 122: Fakhry C, Zhang Q, Nguyen-Tan PF, et al. Human papillomavirus and overall survival after progression of oropharyngeal squamous cell carcinoma. J Clin Oncol 2014; 32: Schöder H, Fury M, Lee N, Kraus D. PET monitoring of therapy response in head and neck squamous cell carcinoma. J Nucl Med 2009; 50(suppl 1):74S 88S 7. Lee JC, Kim JS, Lee JH, et al. F-18 FDG-PET as a routine surveillance tool for the detection of recurrent head and neck squamous cell carcinoma. Oral Oncol 2007; 43: Lell M, Baum U, Greess H, et al. Head and neck tumors: imaging recurrent tumor and post-therapeutic changes with CT and MRI. Eur J Radiol 2000; 33: Evangelista L, Cervino AR, Chondrogiannis S, et al. Comparison between anatomical cross-sectional imaging and 18 F-FDG PET/CT in the staging, restaging, treatment response, and long-term surveillance of squamous cell head and neck cancer: a systematic literature overview. Nucl Med Commun 2014; 35: Subramaniam RM, Truong M, Peller P, Sakai O, Mercier G. Fluorodeoxyglucose-positron-emission tomography imaging of head and neck squamous cell cancer. AJ 2010; 31: Krabbe CA, Pruim J, Dijkstra PU, et al. 18 F-FDG PET as a routine posttreatment surveillance tool in oral and oropharyngeal squamous cell carcinoma: a prospective study. J Nucl Med 2009; 50: Gupta T, Master Z, Kannan S, et al. Diagnostic performance of post-treatment FDG PET or FDG PET/CT imaging in head and neck cancer: a systematic review and meta-analysis. Eur J Nucl Med Mol Imaging 2011; 38: Gao S, Li S, Yang X, Tang Q. 18 FDG PET-CT for distant metastases in patients with recurrent head and neck cancer after definitive treatment: a metaanalysis. Oral Oncol 2014; 50: Kim JW, Roh JL, Kim JS, et al. 18 F-FDG PET/CT surveillance at 3 6 and 12 months for detection of recurrence and second primary cancer in patients with head and neck squamous cell carcinoma. Br J Cancer 2013; 109: Paidpally V, Tahari AK, Lam S, et al. Addition of 18 F-FDG PET/CT to clinical assessment predicts overall survival in HNSCC: a retrospective analysis with for 12 years. J Nucl Med 2013; 54: Nakamura S, Toriihara A, Okochi K, Watanabe H, Shibuya H, Kurabayashi T. Optimal timing of post-treatment [ 18 F]fluorodeoxyglucose-PET/CT for patients with head and neck malignancy. Nucl Med Commun 2013; 34: Ho AS, Tsao GJ, Chen FW, et al. Impact of positron emission tomography/computed tomography surveillance at 12 and 24 months for detecting head and neck cancer recurrence. Cancer 2013; 119: Salaun PY, Abgral R, Querellou S, et al. Does 18 fluoro-fluorodeoxyglucose positron emission tomography improve recurrence detection in patients treated for head and neck squamous cell carcinoma with negative? Head Neck 2007; 29: Kim G, Kim YS, Han EJ, et al. FDG-PET/CT as prognostic factor and surveillance tool for postoperative radiation recurrence in locally advanced head and neck cancer. Radiat Oncol J 2011; 29: Kubota K, Yokoyama J, Yamaguchi K, et al. FDG-PET delayed imaging for the detection of head and neck cancer recurrence after radio-chemotherapy: comparison with MRI/CT. Eur J Nucl Med Mol Imaging 2004; 31: Whiting PF, Rutjes AW, Westwood ME, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011; 155: [No authors]. QUADAS-2: background document. quadas/ migrated/documents/background-doc.pdf. Accessed May 13, Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003; 327: Ng SH, Chan SC, Yen TC, et al. Comprehensive imaging of residual/recurrent nasopharyngeal carcinoma using whole-body MRI at 3 T compared with FDG-PET-CT. Eur Radiol 2010; 20: Kao CH, ChangLai SP, Chieng PU, Yen RF, Yen TC. Detection of recurrent or persistent nasopharyngeal carcinomas after radiotherapy with 18-fluoro-2-deoxyglucose positron emission tomography and comparison with computed tomography. J Clin Oncol 1998; 16: Stokkel MP, Terhaard CH, Hordijk GJ, van Rijk PP. The detection of local recurrent head and neck cancer with fluorine-18 fluorodeoxyglucose dualhead positron emission tomography. Eur J Nucl Med 1999; 26: Li P, Zhuang H, Mozley PD, et al. Evaluation of recurrent squamous cell carcinoma of the head and neck with FDG positron emission tomography. Clin Nucl Med 2001; 26: Tsai MH, Shiau YC, Kao CH, Shen YY, Lin CC, Lee CC. Detection of recurrent nasopharyngeal carcinomas with positron emission tomography using 18-fluoro-2-deoxyglucose in patients with indeterminate magnetic resonance imaging findings after radiotherapy. J Cancer Res Clin Oncol 2002; 128: Wong RJ, Lin DT, Schöder H, et al. Diagnostic and prognostic value of [ 18 F]fluorodeoxyglucose positron emission tomography for recurrent head and neck squamous cell carcinoma. J Clin Oncol 2002; 20: Yen RF, Hung RL, Pan MH, et al. 18-Fluoro-2-deoxyglucose positron emission tomography in detecting residual/recurrent nasopharyngeal carcinomas and comparison with magnetic resonance 638 AJR:205, September 2015