Fikri Selcuk Simsek  ( Department of Nuclear Medicine, Firat University )
Reyhan Koroglu  ( Turkey; Department of Nuclear Medicine, Afyon Kocatepe University, Turkey )
Ferhan Elmali  ( Department of Biostatistics, Izmir Katip Celebi University )
Aylin Comak  ( Turkey; Department of Nuclear Medicine, Inonu University, Turkey )
Hakki Ulutas  ( Turkey; Department of Nuclear Medicine, Inonu University, Turkey )
Tansel Ansal Balci  ( Department of Nuclear Medicine, Firat University )
Muhammed Asik  ( Turkey; Department of Nuclear Medicine, Afyon Kocatepe University, Turkey )
Aysenur Akatli  ( Department of Pathology, Inonu University, Turkey )
Ersoy Kekilli  ( Turkey; Department of Nuclear Medicine, Inonu University, Turkey )
Ali Ozan Oner  ( Turkey; Department of Nuclear Medicine, Afyon Kocatepe University, Turkey )

Objective: To assess whether more accurate mediastinal lymph nodes radiotherapy can be performed with fluorode oxyglu cosepositron emission tomogaphy/computed tomogr ap hy.

Methods: The retrospective study was conducted at Inonu University Medical Faculty, Malatya, Turkey, and Afyon Kocatepe University Medical Faculty, Afyon, Turkey, and comprised record of patients histopathologically diagnosed with non-small cell lung carcinoma and who underwent fluorodeoxyglucose positron emission tomography / computed tomography between January 2013 and December 2016. Surgery and pathology reports of the patients were reviewed. Histopathologically proven malignant and benign lymph nodes were re-identified with fluorodeoxyglucose positron emission tomography / computed tomography imaging. Anatomical and metabolic parameters of lymph nodes were re-assessed by specialists and compared with histopathology reports. Maximum standardised uptake values were used to assess sensitivity, specificity, positive predictive value, and negative predictive values. SPSS 22 was used for data analysis.

Results: The study included 144 mediastinal lymph nodes related to 42 patients who had a mean age of 62.4±9.8 years (range: 41-79 years). In terms of subtypes of the primary squamous cell carcinoma was found in 24(57.2%) patients, adenocarcinoma in 12(27.5%), and other subtypes in 6(15.3%) patients. Of the 144 lymph nodes, 48(33.3%) were metastatic. Sensitivity, specificity, positive predictive value, and negative predictive value were 92.8%, 64.3%, 56.9%, and 94.7%, respectively when maximum standardised uptake value >2.5 was used as the malignancy criterion. When lymph node maximum standardised uptake value / liver standardised uptake value-mean>1.69 was used as the criterion, the sensitivity, specificity, positive predictive value, and negative predictive value were 95.83%, 91.67%, 85.2%, and 97.8%, respectively. When the same values with lymph node >8mm was used as the criterion, the four resultant values were 89.6%, 93.8%, 87.8%, and 94.7%, respectively. When lymph node was replaced with mean attenuation >35 as the criterion, the consequent values were 79.2%, 93.8%, 86.4%, and 90.0%, respectively.

Conclusion: Lymph node maximum standardised uptake value / liver standardised uptake valuemean> 1.69 was associated with higher negative predictive value and more useful positive predictive value compared to maximum standardised uptake value >2.5. When this parameter was used along with short axis or mean attenuation value, there were no significant increase in positive predictive value, but there was a decrease in negative predictive value.

Keywords: Carcinoma, Non-small-cell lung, Radiation therapy, PET scan. (JPMA 70: 29; 2020). https://doi.org/10.5455/JPMA.296815

Introduction

Early stage non-small cell lung carcinoma (NSCLC) is one of the most commonly diagnosed malignant diseases worldwide.1,2 Despite the availability of extensive therapeutic approaches in such cases, very high recurrence / relapse rates cannot be avoided. For example, Pless et al. reported that overall local recurrence (LR) was 60% after surgery and chemo/radiotherapy for NSCLC. It is notable that 30% recurrences occur within the radiotherapy (RT) target region, including mediastinal lymph nodes (MLNs).3 The technique of 18-Fluorodeoxyglucose positron emission tomography / computed tomography (18-FDGPET/ CT ) is widely used for RT planning. For primary tumour RT planning using FDG-PET/CT, the most important problem is mismatch between PET and CT images, which can be solved by the use of 4D FDGPET/ CT.4 On the other hand, maximum standardized uptake value (SUVmax) >2.5 is the currently accepted threshold for NSCLC, and the RT target regions are determined as those with SUVmax 2.0 in FDG-PET/CT. 5This is a reasonable practice considering that normal lung tissue has very little FDG uptake. On the contrary,  accepting SUVmax >2.5 or higher FDG uptake (in comparing MLNs to the mediastinal blood pool [MBP]) for nodal RT planning, which is currently used, cannot seem suitable due to the much higher basal FDG uptake in the mediastinum. Thus, the threshold set for the primary tumour may not be appropriate for MLNs. Many studies have reported extremely low positive predictive value (PPV ) in MLNs with SUVmax >2.5 or higher LN activity compared to the MBP. 6,7Nevertheless, these criteria are currently used because of their high negative predictive value (NPV). The additional parameter of LN>10mm is commonly included in these parameters. 8,9However, this approach causes unnecessary beaming of some non-metastatic LNs and lower RT doses to some malignant ones. Occasionally, endobronchial ultrasound needle aspiration (EBUS-NA) has been used along with FDG-PET/CT to resolve this problem, but this has unexpectedly led to a larger RT field and 20% MLN recurrence. 10Therefore, we need a new practical FDG-PE T/C T parameter for mediastinal nodal RT planning which mayachieve increased PPV without a decrease in NPV. Rogash et al. attained a higher PPV using the LN SUVmax/liver SUVmean ratio in NSCLC patients for mediastinal nodal staging. 11 In addition, Fleching et al. reported significant density differences between malignant and benign MLNs. 12 These results lead to the hypothesis that the use of LN SUVmax/liver SUVmean alone or along with size or density may be more helpful in RT planning for MLNs than the currently used parameters. The current study was planned to assess whether more accurate mediastinal RT planning can be performed with FDG-PET/CT. If more accurate planning can be achieved using a new method, unnecessary irradiation of healthy tissue can be avoided and more effective planning for RT of malignant MLNs can be achieved.

Materials and Methods 

The retrospective study was conducted at Inonu University Medical Faculty, Malatya, Turkey, and Afyon Kocatepe University Medical Faculty, Afyon, Turkey, and comprised record of patients histopathologically diagnosed with NSCLC and who underwent FDG-PET/CT between January 2013 and December 2016. Those excluded were records of patients with previously diagnosed malignancy, chemotherapy and / or RT before FDG-PET/CT, locally advanced / advanced stage NSCLC per the 8th edition of the tumour, nodes, metastases (TNM) system, 2 or invasive sampling from MLNs before FDG-PET/CT due to possible inflammation. Surgery and pathology reports of the selected patients were re-examined and histopathologically proven malignant or benign LNs were re-identified carefully in FDG-PET/C T images. Anatomical and metabolic parameters of LNs were assessed by nuclear medicine specialists and MLNs from the patients were evaluated histopathologically. For PET / CT, a 0.1 mCi/kg FDG dose was administered to patients intravenously (IV) after at least six hours of fasting. Subsequently, patients were hydrated with 500ml water orally while they were resting in a quiet, dark room. Between injection and imaging, there was a waiting period of 60±5 minutes. PET/CT scanning was performed using a Siemens Biograph mCTS (20)-3R (Knoxville, TN, USA) or a Siemens Biograph 6 True Point (Knoxville, TN, USA). Following CT imaging from vertex to mid-thigh, PET scanning of the same regions was performed in 7-8 bed positions. An ordered subset expectation maximisation algorithm was used for reconstruction. The regions of interest (ROIs) were drawn manually for primary tumours and MLNs. ROIs for liver SUVmean were also drawn manually for normal liver tissues 2.5 x 2.5cm in size. Non-attenuation-corrected images were used to examine calcified LNs . In surgical procedure, mediastinal and hilar LN stations were explored based on the clinical stage of the lesion using mediastinoscopy or video-assisted thoracoscopic surgery (VATS) / thoracotomy. All LNs with SUVmax >2.5 and short axis 1.0cm or any suspected LNs were removed as much as possible. They were dissected out along with their capsules. EBUS-NA was performed in patients not sampled with the two above-mentioned techniques. Patients, in whom EBUS-NA did not provide exact diagnoses, were surgically sampled. For histopathological examination, formalin-fixed paraffin-embedded tissues were cut into 4-5m thick sections and kept in 10% formalin for at least 24 hours. The sections were stained with Haematoxylin-and-Eosin (H&E) and examined under a light microscope by pathologists. This evaluation was accepted as the gold standard. Data was analysed using SPSS 22. Frequencies and percentages, mean±standard deviation (SD), and median and interquartile ranges (IQRs) were determined. Normal distribution of numerical variables was evaluated using Shapiro Wilk normality test and Q-Q graphs. Inter-group comparisons were performed using independent twosample t-tests for variables in normal distribution and using Mann-Whitney U test for variables in non-normal distribution. Associations between numerical variables were analysed using Spearman correlation analysis. Receiver operating characteristic (ROC) curve analysis was used to evaluate diagnostic tests and to calculate the cut-off values. Associations between categorical variables was calculated using Fisher's exact test. P<0.05 was considered statistically significant.

Results

Of the 496 records reviewed, 42(8.5%) comprised the study sample. They related to patients who had a mean age of 62.4±9.8 years (range: 41-79 years). In terms of subtypes of the primary squamous cell carcinoma was found in 24(57.2%) patients, adenocarcinoma in 12(27.5%), and other subtypes in 6(15.3%) patients. Of the 144 lymph nodes studied, 48(33.3%) were metastatic LNs. The differences in C T short axis and mean attenuation values between malignant and benign LNswere statistically significant (p<0.05). Mean size and mean attenuation values were 14.9±4.6 mm and 41.8±12.1 respectively in malignant LNs and 8.0±4.2 mm and 35.2±13.5 respectively in benign LNs. When short axis 10 mm was used as the malignancy criterion, the sensitivity, specificity, PPV, NPV and accuracy were 84.8%, 81.3%, 65.1%, 92.9% and 82.3%, respectively. ROC analysis showed that the cut-off value for short axis of LNs was >8mm. With this value, sensitivity, specificity, PPV and NPV were 93.8%, 71.9%, 62.5% and 95.8% respectively and area under the ROC curve was 89.4. We found >35 as the cut-off CT mean attenuation value for LNs, with the sensitivity, specificity, PPV and NPV being 81.3%, 55.2%, 47.6% and 85.5% respectively and area under the ROC curve was 66.5. The mean SUVmax values for malignant LNs and benign LNs were 10.9±5.9 and 2.41±1.2, respectively, which were statistically significant (p<0.05). Using SUVmax >2.5, the sensitivity, specificity, PPV and NPV were 92.8%, 64.3%, 56.9% and 94.7% respectively. The cut-off value for LN SUVmax/liver SUVmean was >1.69 and with this cut-off, the sensitivity, specificity, PPV and NPV were 95.8%, 91.7%, 85.2%, and 97.8% respectively (Table 1).



When the parameter LN SUVmax/liver SUVmean >1.69 was used along with LN >8mm as the malignancy criterion, the sensitivity, specificity, PPV and NPV were 89.6%, 93.8%, 87.8% and 94.7% respectively. When LN SUVmax/liver SUVmean >1.69 along with mean attenuation >35 was used as the malignancy criterion, the sensitivity, specificity, PPV and NPV were 79.2%, 93.8%, 86.4% and 90.0% respectively ( Table 2).



Discussion

To the best of our knowledge, the present study is the first one to focus on the PPV of different parameters, like LN SUVmax/liver SUVmean alone or along with size or density, in RT planning for MLNs. NSCLC is one of the most commonly diagnosed malignancies worldwide. Although extensive treatment approaches are available, it is the leading cause of cancerrelated mortality. Among the therapy options, RT can  be used in early, local advanced and advanced stagepatients. It has an important role in advanced stage NSCLC in palliative therapy and can also be applied as a part of neoadjuvant / adjuvant therapy in operable cases. In addition, it can be performed in patients who cannot undergo surgery due to medical and / or personal causes.3-5,8-10 Despite aggressive therapy during early stages, the recurrence rate of NSCLC is very high. Melloni et al. reported 20-40% recurrence even in stage I patients. 13Recurrence may occur at distant or local sites. Lou et al. found 26% local recurrence or 30% local recurrence along with distant recurrence in early stage NSCLC patients. 14 In some cases, different RT approaches were used to address the problem, but they were unsuccessful and led to more side effects. 15,16The main aim in cancer therapy is to preserve normal tissues while eliminating all malignant tissues. This goal has not yet been achieved in NSCLC. RT planning using FDG-PET/CT provided better results compared to CT-based planning in studies over the last decade. 17 However, the results are far from sufficient and have included MLNs recurrences and relapses. Billiet et al. reported approximately 37% local recurrence after postoperative RT planned using FDGPET/ CT with some LN regions showing recurrence rates as high as 22%. An interesting point to address in these reports is whether recurrence was in the mediastinal RT planning region or outside this area.10 Nygard et al. planned RT based on FDG-PET/CT and found 14% LN recurrences outside the RT region. 18 Both the above recent studies demonstrated no sufficient decrease in nodal recurrence using classical FDG-PET/CT parameters. In planning for RT of primary tumours using FDG-PET/CT, a consensus has emerged regarding the SUVmax cut-off value and a strategy to overcome mismatch between CT and PET. 4,5 On the other hand, no such consensus exists for MLNs. Usefulness of metabolic parameters, suitable cut-off value, and the utility of combined use

of tomography and PET parameters remain unclear.  Because of this ambiguity, SUVmax >2.5 or increased LNmetabolic activity compared to the MBP continues to be used as a positivity criterion due to the high NPV of these parameters. Some studies included the additional  parameter of LNs >10mm but this resulted in a widenedRT region. ^4-6,10,16 The important point is that all the above parameters have a low PPV for MLN evaluation. Mallorie et al. reported 46.8% PPV and 97.7% NPV of the parameter LN activity>MBP activity.19 Li et al. used SUVmax >2.5 as an indicator of malignancy in mediastinal nodes and found sensitivity, specificity, PPV, NPV and accuracy of 74.2%, 73.2%, 54.4%, 86.8%, and 73.5%, respectively. 20 A meta-analysis yielded values different from the classical criteria; however, the findings were also insufficient.6 Thus, planning for RT of MLNs is quite challenging for radiation oncologists. Some studies have used EBUS-NA and FDG-PET/CT together. Peeters et al.used visual interpretation of FDG-PET/CT and EBUS-NA together and found that FDG-PET/CT-negative and EBUSNA- negative LNs can be excluded from the RT region. However, if an LN is EBUS-NA-positive or has higher FDG uptake than the MBP, it should be included except in the case of bilateral symmetrical FDG-PET/CT positivity. 9 This caused widening of the planned target volume in MLNs and probably unnecessary beaming for some nodes due to low PPV of this metabolic parameter. A more helpful metabolic parameter with higher PPV for mediastinal nodal RT planning in NSCLC can enable more specific beaming, may be leading to decreased mediastinal recurrence / relapse rates and radiaation side effects. To address the problems of low specificity and PPV, some studies used LN SUVmax/liver SUVmean ratios. Rogash et al. found sensitivity, specificity, PPV, NPV and accuracyof 80.6%, 88.2%, 67.4%, 93.8%, and 86.5%, respectively, using LN SUVmax/liver SUVmean >1.7.11 Nguyen et al. reported sensitivity, specificity and accuracy of 81.6%, 97.6%, and 86.7% respectively using LN SUVmax/liver SUVmean 2.31.21 The current first examined LN SUVmax>2.5, which is a standard semi-quantitative method used by most specialists in mediastinal nodal RT planning. Similar to other studies 10,11 we observed very low PPV (56.9%) and specificity (64.3%), with 92.8% sensitivity and 94.7% NPV. However, when we examined LN SUVmax/liver SUVmean using ROC analysis, the sensitivity, specificity, PPV and NPV were 95.8%, 91.7%, 85.2%, and 97.8%, respectively, with a cut-off >1.69. These were significantly more accurate compared to SUVmax >2.5 (Figure).



Interestingly, this value is very similar to that reported by Rogash et al. 11  On the other hand, Nguyen et al. calculated a cut-off value quite different compared to ours, using the same metabolic parameter. Different patient selection criteria and the high number of SCC patients in our study, or the high number of large cell carcinoma patients in the other study21  may have caused this difference. Although insufficiency of single CT parameters is known, use of these parameters along with metabolic parameters must be investigated for improving accuracy. It is important to note that in our study, the cut-off value of >8mm for the LN short axis provided more accurate results compared to >10mm, which is used currently, with ROC analysis. In addition, short axis >8mm provided more reliable results compared to SUVmax >2.5. A similar finding was reported by Lee et al. 22  In terms of the use of the short axis in combination with metabolic parameters, studies mostly focused on NPV and sensitivity but not PPV. 6,20  In contrast to most studies, Li et al. investigated specificity in particular and found that if attenuation <70HU and SUVmax >2.5 were accepted as malignancy criteria in non-calcified LNs, the sensitivity was 68% and specificity was 95%. If SUVmax >3.5 was used with the same CT parameter, the sensitivity was only 13% with 94% specificity. 23  Lee at al. reported that if SUVmax >4.0 or SUVmax 2.0-4.0 along with LN mean attenuation values of 25-45 were used, the sensitivity, specificity and accuracy were 88.3%, 82.6%, and 86.0% respectively. 24  We used LN SUVmax/liver SUVmean >1.69 along with LN size >8mm or LN mean attenuation value >35 separately. Using this approach, the former parameter's sensitivity was 6.2% lower and NPV 3.1% lower than that using LN SUVmax/liver SUVmean >1.69 alone, whereas increase in PPV was 2.6%. The latter parameter caused more dramatic decreases (16.6% in sensitivity and 7.8% in specificity), with only a 1.2% increase in PPV compared to LN SUVmax/liver SUVmean >1.69 alone. In the final analysis, use of LN SUVmax/liver SUVmean >1.69 alone appeared more appropriate compared to other parameters in mediastinal nodal RT planning. However, the findings of the current study must be validated through prospective studies with larger patient series.

 

Conclusion

LN SUVmax/liver SUVmean >1.69 had a higher NPV than the currently used parameters, and had a more useful PPV as well. Use of this parameter along with short axis or mean attenuation value did not result in a significant increase in PPV, but a decrease in NPV was seen. This parameter may be more beneficial in mediastinal nodal RT planning than the classical ones, enabling more selective RT and providing a strategy to decrease nodal

recurrences.

Disclaimer: None.

Conflict of Interest: None.

Source of Funding: None.

References

 

1. Allemani C, Matsuda T, Di Carlo V, Harewood R, Matz M, Nikši? M, et al. CONCORD Working Group. Global surveillance of trends in cancer survival 2000-14 (CONCORD-3): analysis of individual records for 37?513?025 patients diagnosed with one of 18 cancers from 322 population-based registries in 71 countries. Lancet 2018; 391: 1023-75.
2. Goldstraw P, Chansky K, Crowley J, Rami-Porta R, Asamura H, Eberhardt WE, et al. The IASLC Lung Cancer Staging Project: Proposals for Revision of the TNM Stage Groupings in the Forthcoming (Eighth) Edition of the TNM Classification for Lung Cancer. J Thorac Oncol 2016; 11:39-51.

3. Pless M, Stupp R, Ris HB, Stahel RA, Weder W, Thierstein S, et al. Induction chemoradiation in stage IIIA/N2 non-small-cell lung cancer: a phase 3 randomised trial. Lancet 2015; 386: 1049-56.

4. Callahan J, Kron T, Siva S, Simoens N, Edgar A, Everitt S, et al. Geographic miss of lung tumours due to respiratory motion: a comparison of 3D vs 4D PET/CT defined target volumes. Radiat Oncol 2014; 9: 291.

5. Zhang Y, Li J, Dun Y, Wang W, Li F, Shao Q, et al. Comparison of biological target volume metrics based on FDG PET-CT and 4DCT for primary non-small-cell lung cancer. Oncotarget 2017; 8: 79629- 35

6. Pak K, Park S, Cheon GG, Kang KW, Kim IJ, Lee DS, et al. Update on nodal staging in non-small cell lung cancer with integrated positron emission tomography/computed tomography: a meta-analysis. Ann Nucl Med 2015; 29:409-19.

7. Lee AY, Choi SJ, Jung KP, Park JS, Lee SM, Bae SK. Characteristics of Metastatic Mediastinal Lymph Nodes of Non-Small Cell Lung Cancer on Preoperative F-18 FDG PET/CT. Nucl Med Mol Imaging 2014; 48: 41-6.

8. Ding XP, Zhang J, Li BS, Li HS, Wang ZT, Yi Y, et al. Feasibility of shrinking field radiation therapy through 18F-FDG PET/CT after 40Gy for stage III non-small cell lung cancers. Asian Pac J Cancer Prev 2012; 13: 319-23.

9. Peeters ST, Dooms C, Van Baardwijk A, Dingemans AM, Martinussen H, Vansteenkiste J et al. Selective mediastinal node irradiation in non-small cell lung cancer in the IMRT/VMAT era: How to use E(B)US-NA information in addition to PET-CT for delineation? Radiother Oncol 2016; 120: 273-8.

10. Billiet C, De Ruysscher D, Peeters S, Decaluwé H, Vansteenkiste J, Dooms C, et al. Patterns of Locoregional Relapses in Patients with Contemporarily Staged Stage III-N2 NSCLC Treated with Induction Chemotherapy and Resection: Implications for Postoperative Radiotherapy Target Volumes. J Thorac Oncol 2016; 11: 1538-49.

11. Rogasch JM, Steffen IG, Riedel S, Apostolova I, Wertzel H, Achenbach HJ, et al. Dual time point imaging for F18-FDG-PET/CT does not improve the accuracy of nodal staging in non-small cell lung cancer Patients. Eur Radiol 2016; 26: 2808-18.

12. Flechsig P, Kratochwil C, Schwartz LH, Rath D, Moltz J, Antoch G, et al. Quantitative volumetric CT-histogram analysis in N-staging of 18F-FDG-equivocal patients with lung cancer. J Nucl Med 2014; 55: 559-64.

13. Melloni G, Gajate AM, Sestini S, Gallivanone F, Bandiera A, Landoni C, et al. New positron emission tomography derived parameters as predictive factors for recurrence in resected stage I non-small cell lung cancer. Eur J Surg Oncol 2013; 39: 1254-61.

14. Lou F, Huang J, Sima CS, Dycoco J, Rusch V, Bach PB, et al. Patterns of recurrence and second primary lung cancer in early-stage lung cancer survivors followed with routine computed tomography surveillance. J Thorac Cardiovasc Surg 2013; 145: 75-81; discussion 81-2.

15. Machtay M, Bae K, Movsas B, Paulus R, Gore EM, Komaki R, et al. Higher biologically effective dose of radiotherapy is associated with improved outcomes for locally advanced non-small cell lung carcinoma treated with chemoradiation: an analysis of the Radiation Therapy Oncology Group. Int J Radiat Oncol Biol Phys 2012; 82,1:425-34.

16. Wanet M, Delor A, Hanin FX, Ghaye B, Van Maanen A, Remouchamps V, et al. An individualized radiation dose escalation trial in non-small cell lung cancer based on FDG-PET imaging. Strahlenther Onkol 2017; 193: 812-22.

17. van Loon J, van Baardwijk A, Boersma L, Ollers M, Lambin P, De Ruysscher D. Therapeutic implications of molecular imaging with PET in the combined modality treatment of lung cancer. Cancer Treat Rev 2011; 37: 331-43.

18. Nygård L, Vogelius IR, Fischer BM, Klausen TL, Langer SW, Lonsdale MN, et al. Early lesion-specific 18F-FDG PET response to chemotherapy predicts time to lesion progression in locally advanced non-small cell lung cancer. Radiother Oncol 2016; 118: 460-4.

19. Mallorie A, Goldring J, Patel A, Lim E, Wagner T. Assessment of nodal involvement in non-small-cell lung cancer with 18F-FDGPET/ CT: mediastinal blood pool cut-off has the highest sensitivity and tumour SUVmax/2 has the highest specificity. Nucl Med Commun 2017; 38: 715-9.

20. Li S, Zheng Q, Ma Y, Wang Y, Feng Y, Zhao B, et al. Implications of False Negative and False Positive Diagnosis in Lymph Node Staging of NSCLC by Means of 18F-FDG PET/CT. PLoS One 2013; 25: e78552.

21. Nguyen P, Bhatt M, Bashirzadeh F, Hundloe J, Ware R, Fielding D, et al. Comparison of objective criteria and expert visual interpretation to classify benign and malignant hilar and mediastinal nodes on 18-F FDG PET/CT. Respirology 2015; 20:129-37.
22. Lee SM, Park CM, Paeng JC, Im HJ, Goo JM, Lee HJ et al. Accuracy and predictive features of FDG-PET/CT and CT for diagnosis of lymph node metastasis of T1 non-small-cell lung cancer manifesting as a sub solid nodule Eur Radiol 2012; 22: 1556- 63.

23. Li X, Zhang H, Xing L, Ma H, Xie P, Zhang L et al. Mediastinal lymph nodes staging by 18F-FDG PET/CT for early stage nonsmall cell lung cancer: a multicenter study. Radiother Oncol J Eur Soc Ther Radiol Oncol 2012; 102: 246-50.

24. Lee JW, Kim EY, Kim DJ, Lee JH, Kang WJ, Lee JD, et al. The diagnostic ability of 18F-FDG PET/CT for mediastinal lymph node staging using 18F-FDG uptake and volumetric CT histogram analysis in non-small cell lung cancer. Eur Radiol 2016; 26: 4515-23.