| Study | Reason for exclusion |
|---|---|
| Agraval 2012 | 2‐by‐2 table could not be extracted for patient‐level N‐staging data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Ahmed 2010 | Conference abstract. 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 15 November 2012 |
| Akpinar 2013 | Index test scanning employed an ECAT Accel scanner (Siemens, Erlangen, Germany), which is a dedicated PET scanner, that is, not an integrated PET/CT scanner |
| Al‐Ibraheem 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Al‐Sarraf 2008 | The unit of analysis was lymph node. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Allen‐Auerbach 2006 | 29/142 included participants had NSCLC. The remainder had other types of cancer. 2‐by‐2 table could not be extracted for this subgroup. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012. Author responded that study was not set up for assessing the accuracy of PET‐CT for staging of NSCLC. The aim of the study was solely to determine the incidence of missed pulmonary micronodules on PET/CT studies acquired during shallow breathing. Thus, the data relevant for this review were not part of this study |
| An 2008 | The unit of analysis was lymph node. 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Antoch 2003 | N2 status of the participants could not be ascertained as participants were classified according to overall stage with no individual breakdown of T‐, N‐, and M status. Emailed for per‐patient‐based‐data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Aquino 2003 | FDG‐PET and CT performed on different scanners within a 1‐month time interval and subsequently coregistered |
| BalciI 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Beyer 2010 | Reference standard consisted of a combination of FDG uptake in lymph nodes, findings from either histological examinations , FDG‐PET/CT, or both, and clinical follow‐up examinations for a minimum of 2 months |
| Bhatt 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Booth 2009 | Conference abstract. 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Boulougouri 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Bryant 2006b | 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0‐1 v N2 and N3 on 1 November 2012 |
| Carrillo 2012 | 2‐by‐2 table could not be extracted for patient level N0‐1 v N2 and N3 data. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013. Correspondance with the author clarified that N1 and N2 disease were treated as N‐positive as a whole and not distinguished between |
| Cerfolio 2004 | 13 participants with pathological N0 disease were over‐staged by FDG‐PET/CT. Not enough information was provided to ascertain which N‐stage these participants were assigned to by FDG‐PET/CT. It was therefore not possible to know how many of these participants were true negatives (if staged by FDG‐PET/CT as N1 disease) and how many of these participants were false positives (if staged as N2 or N3 disease by FDG‐PET/CT). Similarly, 2 participants with pathological N3 disease were under‐staged by FDG‐PET/CT. Not enough information was provided to ascertain which N‐stage these participants were staged as by FDG‐PET/CT, and it is therefore impossible to know whether these participants should be classed as true positives (if staged as N2 disease by FDG‐PET/CT) or false negatives (if staged as N0 or N1 disease by FDG‐PET/CT). Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Cerfolio 2006 | Included only participants who were clinically N0 and N1 and excluded all participants who were clinically N2, N3, and M1 even if those test results later turned out to be false positive, which meant no false positive results would be possible in the included data set |
| Cerfolio 2007 | Unit of analysis = node. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Cerfolio 2008 | Data can only be verified for 14 to 26/166 participants. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
| Cetinkaya 2011 | Unclear if all participants received PET‐CT. 2‐by‐2 could not be extracted. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
| Ceylan 2012 | 2‐by‐2 could not be extracted. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
| Chiba 2010 | No pathological reference standard |
| Colville 2013 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Cömert 2012 | 2‐by‐2 could not be extracted. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
| Delgado‐Bolton 2010 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Duan 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Faber 2011 | Conference abstract. 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Flechsig 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Gregory 2012 | 61/168 participants received the reference standard |
| Gómez‐Caro 2012 | Included only participants who were clinical stage 1 and excluded all participants who were clinical stage IIA and above, which meant no false positive results would be possible in the included data set |
| Günlüoğlu 2010 | 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Günlüoğlu 2011b | Published in Turkish. Unsure if it is PET or PET‐CT. 2‐by‐2 table could not be extracted. Emailed author for clarification and per‐patient‐based data for N0 and N1 v N2 and N3 on 18 March 2013 |
| Halpern 2005 | 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Hong 2010 | Conference abstract. 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Hu 2008 | 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Huang 2012 | Included only PET‐CT‐negative participants, so no TP or FP possible |
| Kasai 2010 | 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Kelly 2006 | 18/49 participants did not receive reference standard |
| Kim 2011 | Included only participants who were staged N0 and N1 on both CT and PET‐CT, and excluded all participants who were staged N2 and N3 or M1 on CT or PET‐CT, which meant no false positive results would be possible in the included data set |
| Kim 2012 | 2‐by‐2 could not be extracted. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
| Kim 2012a | 9/69 participants received the reference standard |
| Kim 2012b | Included only PET/CT‐negative participants, so no possibility of TPs or FPs |
| Kommata 2011 | 337/401 participants did not receive reference standard |
| Krueger 2006 | Conference abstract. 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 5 November 2012 |
| Lapinska 2011 | Conference abstract. 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Lardinois 2003 | 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Lasnon 2012 | 25/71 participants received reference standard |
| Lebioda 2013 | Only PET‐CT N2‐negative nodes were investigated, thus, no TPs or FPs possible |
| Lee 2004 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013, but received delivery failure report on email |
| Lee 2008 | Unit of analysis = node. 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Lee 2009b | During the study period 63 newly diagnosed NSCLC participants underwent both CT and integrated PET‐CT for clinical staging. Of those, 43 participants underwent surgical procedures (including mediastinoscopy), which allowed pathologic evaluation of the mediastinal nodes, and only these 43 participants were included in the study, of whom data can only be ascertained for 37 |
| Li 2009 | 2‐by‐2 table could not be extracted for patient‐level N‐staging data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Li 2011 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Li 2012b | 2‐by‐2 could not be extracted. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
| Li 2012c | Only PET‐CT positive nodes were investigated, thus, no TNs or FNs possible |
| Lin 2012 | 2‐by‐2 could not be extracted. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
| Liu 2009 | Unit of analysis = node. 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Low 2006 | 17/41 participants received the gold standard |
| Ma 2011 | 2‐by‐2 table could not be extracted for patient‐level N‐staging data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Maeda 2009 | Included only participants who were clinical stage IA, which meant no false positive results would be possible in the included data set |
| Mariam 2009 | Conference abstract. 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 5 November 2012 |
| Meduoye 2009 | Not integrated FDG‐PET/CT, but FDG‐PET and CT performed separately |
| Mendez 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Mi 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Moodie 2009 | No reference standard |
| Moreno Garcia 2009 | Conference abstract. 2‐by‐2 table could not be extracted. Could not find author contact details to request per‐patient‐based data for N0 and N1 v N2 and N3 |
| Morikawa 2011 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Nakajima 2011 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Nakamura 2008 | Reference test consisted of clinical course or pathology findings in biopsy specimens from informative sites for non‐surgical cases (N = 30/50) |
| Nomori 2008 | 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2‐3 on 1 November 2012 |
| Ohno 2007 | N0 disease classified as test negative and N1, N2, and N3 disease classified as test positives. Not enough information was provided to permit reclassification of N1 disease to test negatives. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Ozkan 2010 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Pauls 2012 | None of the participants received the reference standard |
| Pfannenberg 2007 | Population consists of 50 participants referred for PET/CT for primary tumour staging before surgery (N = 4), neoadjuvant radiochemotherapy (N = 24), definitive radiotherapy (N = 8), or palliative radiotherapy (N = 5). The remaining 9 participants were referred for restaging during follow‐up because of suspected local or distant recurrence Reference standard consisted of surgical staging for 35 participants and follow up for 15 participants |
| Pozo‐Rodriguez 2005 | Not integrated FDG‐PET/CT, but FDG‐PET and contrast‐enhanced CT performed separately on different scanners and not coregistered |
| Prévost 2009 | 2‐by‐2 table could not be extracted for the 44 participants with pathological staging. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Quaia 2008 | N0 disease classified as test negative and N1, N2, and N3 disease classified as test positives. Not enough information was provided to permit reclassification of N1 disease to test negatives. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Salman 2009 | Conference abstract. 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Schiavariello 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Schreyögg 2010 | 77/172 participants received reference standard |
| Schwenzer 2012 | No reference standard |
| Shim 2005 | Retrospective study: unit of analysis = node Prospective study: unit of analysis = node. Overall staging results provided for stages I, II, III, and IV, and 2‐by‐2 table for N‐staging with participant as unit of analysis could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Sit 2010 | 2‐by‐2 table could not be extracted for patient‐level N‐staging data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Sivrikoz 2010 | Only PET‐CT‐negative nodes were investigated, thus, no TPs or FPs possible |
| Sivrikoz 2012 | 2‐by‐2 could not be extracted for all participants. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
| Steinert 2010 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Sánchez Sánchez 2011 | 14/34 participants did not received pathological confirmation of PET/CT results |
| Tamura 2012 | 2‐by‐2 could not be extracted. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
| Tsutani 2012 | Included only PET/CT‐negative participants, i.e., no chance of TPs or FPs |
| Vaz 2012 | 2‐by‐2 could not be extracted. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25/2/13 |
| Ventura 2010 | Unit of analysis = node. 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012, but received delivery failure report. Re‐sent on 5 November 2012 with the same result. We could not find any other contact details for the corresponding author |
| Wang 2012 | 2‐by‐2 table could not be extracted for patient‐level N‐staging data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Wiese 2012 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 using the "contact us" webform on www.leading‐medicine‐guide.ch/en/spitzenmediziner/kontakt/id/2492 |
| Wu 2010 | 2‐by‐2 table could not be extracted for patient‐level N‐staging data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 13 June 2013 |
| Yi 2007 | The participants appear to be retrospectively recruited from the same institution as Shin 2008 during the period of July 2003 until June 2005. Shin 2008 recruited (also retrospectively) their participants during the period September 2003 until July 2006, which means the samples were largely overlapping. We retained Shin Shin 2008 because this study has the larger sample of the 2 studies |
| Yi 2011 | Abstract. Unit of analysis = node. 2‐by‐2 table could not be extracted for patient‐level N‐staging data. Emailed via Kyung Soo Lee for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Yi 2013 | 2‐by‐2 table could not be extracted for patient‐level data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 10 July 2013 |
| Yu 2007 | Paper primarily about diagnosis, not staging, with 34 out of 104 participants having pathological diagnosis and 70 out of 104 participants having clinical diagnosis. 18 of the 104 participants had benign lesions, and 86 participants had lung cancer. 12 of 25 lung cancer participants with no metastasis on FDG‐PET/CT received surgery |
| Zhang 2006 | 2‐by‐2 table could not be extracted. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012, but received delivery failure report. Re‐sent on 7 November 2012 with the same result. We could not find any other contact details for the corresponding author |
| Zhang 2012 | 2‐by‐2 table could not be extracted for patient‐level N‐staging data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Zsiray 2009 | Hungarian. Describes their experience and there appears to be no consistent reference standard, and no possibility of extracting a 2‐by‐2 with clarity that it is N0 and N1 v N2 and N3. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 1 November 2012 |
| Zsiray 2011 | Not pathological reference standard |
| Zsiray 2012 | 2‐by‐2 table could not be extracted for patient‐level N‐staging data. Emailed for per‐patient‐based data for N0 and N1 v N2 and N3 on 17 June 2013 |
| Öztaş 2012 | 2‐by‐2 could not be extracted. Emailed author for per‐patient‐based data for N0 and N1 v N2 and N3 on 25 February 2013 |
FDG = (¹⁸F)‐2‐fluoro‐deoxy‐D‐glucose. FDG‐PET/CT = (¹⁸F)‐2‐fluoro‐deoxy‐D‐glucose positron emission tomography–computed tomography. FPs = false positives. NSCLC = non‐small cell lung cancer. TPs = true positives. v = versus.