Statistical analysis of 18F-fluorodeoxyglucose positron-emission tomography/computed tomography ground-glass nodule findings

Introduction

18F-fluorodeoxyglucose-positron emission tomography/computed tomography (18F-FDG-PET/CT) is widely used to make qualitative diagnoses. When treating lung cancers, for example, it is used to determine the malignancy of lesions in the lungs and lymph nodes. However, PET images generally have poorer resolution than CT images and scans of lung tissue are particularly affected by respiratory rhythms. These factors are considered to limit PET's usefulness in diagnosing small-diameter lesions such as ground-glass nodules (GGNs (1–3). However, few publications have examined PET/CT's utility when examining GGNs alone (4,5). Moreover, it has not been established definitively whether this imaging modality is useful for certain kinds of GGNs and how such data should be evaluated. Well-differentiated adenocarcinomas are among the types of GGN lesions identifiable on CT. However, their small diameter makes it difficult to collect pathological specimens in many cases (for example, by means of bronchoscopy or CT-guided needle lung biopsy). This raises an important question regarding the utility of PET/CT for determining the benignity or malignancy of GGNs. In this study, we examined PET/CT's clinical utility for the diagnosis of GGN lesions by comparing preoperative CT and PET/CT findings of patients who underwent surgery at Japanese Red Cross Okayama Hospital and who were diagnosed with lung cancer based on histological findings.

Patients and methods

Records of patients diagnosed with lung cancer following pulmonary resection at Japanese Red Cross Okayama Hospital between January 2010 and December 2014 were reviewed retrospectively. Only patients who underwent PET/CT and whose preoperative CT findings indicated GGNs were analyzed.

All patients were imaged using an Aquilion 64 CT scanner at Japanese Red Cross Okayama Hospital (Toshiba Medical Systems, Otawara, Japan). The scan settings were as follows: slice dimensions = 512×512 pixels, slice thickness = 1.0 mm, scanning interval = 0.8 mm, tube voltage =120 mA (with automatic tube voltage modulation), and pitch factor = 0.844. GGNs were evaluated on horizontal chest CT images in terms of total diameter (the long axis of the lesion), solid-part diameter (the long axis of the hyperechoic, low-contrast part, as determined visually), and solid-part ratio (the solid-part diameter/total diameter) (Fig. 1). If multiple GGNs were observed simultaneously in a slice, only the lesion with the largest total diameter was analyzed.

Figure 1. CT findings (Part-solid GGN). Symbols: a, Total diameter; b, Solid-part diameter; b/a, Solid-part ratio.

PET/CT scanning was performed at Okayama Diagnostic Imaging Center and Okayama Kyokuto Hospital. At the former, patients were scanned using a lutetium oxyorthosilicate (LSO)-based Biograph Sensation 16 PET/CT scanner with manufacturer-recommended settings (Siemens, Munchen, Germany). FDG (3.7 MBq/kg) was administered intravenously after the patient had fasted for ≥5 h; scanning began 90 min thereafter. At the latter, patients were scanned using a Discovery LS PET/CT system with manufacturer-recommended settings (General Electric Company, Boston, MA, USA). FDG (6 MBq/kg) was administered intravenously after the patient had fasted for ≥4 h; scanning began 50 and 120 min thereafter for early- and delayed-phase images, respectively. Only early-phase images were used in the analysis. Both facilities conduct daily inspections for PET/CT quality control. Various studies have been published on the evaluation of PET findings. This study employed the maximum standardized uptake value (SUVmax), a commonly used metric in Japan at present. Pulmonary nodules were judged to have clinically important FDG uptake when the SUVmax was ≥2.5 (2,6).

The statistical analysis was performed using EZR software version 1.36 (Saitama Medical Center, Jichi Medical University, Saitama, Japan) (7). In the statistical analysis, an SUVmax too low to evaluate was treated as SUVmax=0.0. p≤0.05 was considered to indicate a statistically significant difference in two-group comparisons. Correlations were evaluated using the Pearson product-moment correlation coefficient; r≥0.4 was considered to indicate a moderate correlation. Fischer's exact test was used to analyze frequency distributions. The t-test was used to determine whether differences in the means of two sets of samples were significant.

All lung cancer diagnoses were corroborated by the pathological histology of the biopsied specimens. Original histology-based diagnoses were based on the General Rules for the Clinical and Pathological Classification of Lung Cancer of the Japan Lung Cancer Society (8th edition) and the TNM staging system of the International Association for the Study of Lung Cancer (8th edition) (8). The same tissue specimens were re-examined and classified in the present study using the World Health Organization Classification of Tumors of the Lung, Pleura, Thymus, and Heart (4th edition) (9).

Patients' clinical characteristics, CT findings (total diameter, solid-part diameter, and solid-part ratio), pathological diagnosis, and their relationships with the SUVmax were analyzed statistically. This study was conducted with the approval of the Ethics Committee of Japanese Red Cross Okayama Hospital.

Results

In total, 66 patients who were diagnosed with lung cancer following pulmonary resection at Japanese Red Cross Okayama Hospital between January 2010 and December 2014 were analyzed. All had undergone PET/CT and had preoperative CT findings indicating GGNs. The subjects ranged in age from 47-86 years (median, 69 years) and consisted of 28 men and 38 women. All were diagnosed with lung adenocarcinoma; lymph node/distant metastasis was not observed in any case. The pathologic tumor (pT) factors were 1a, 1b, 1c, 2a, and 2b in 5, 30, 21, 9, and 1 case(s), respectively; the pathologic stage (pStage) classifications were IA1, IA2, IA3, IB, and IIA in 5, 30, 21, 9, and 1 case(s), respectively. Visceral pleural infiltration beyond the elastic layer was observed in one case but did not reach the visceral pleural membrane (i.e., stage pl1). The total diameters of the GGNs according to CT findings ranged from 7.0-41.13 mm (median, 19.43 mm); the solid-part diameters ranged from 0.0-23.23 mm (median, 4.55 mm), and the solid-part ratios ranged from 0-77% (median, 20%). A total of 22 lesions were diagnosed as pure GGNs. The SUVmax ranged from a value too low to be evaluated to a maximum of 3.9 (median, 1.0). The SUVmax ranged from 1.00-1.49 in 34 patients (51.5%), 1.50-1.99 in 19 (28.8%), 2.00-2.49 in 11, (16.7%), and ≥2.5 in 6 (9.1%). The histopathological classifications of the adenocarcinoma subtype were adenocarcinoma in situ (AIS) in 17 patients, minimally invasive adenocarcinoma (MIA) in 15, lepidic-predominant adenocarcinoma (LPA) in 32, and papillary-predominant adenocarcinoma (PPA) in 2 (10) (Table I). The SUVmax correlated with each CT metric as follows: r=0.513 for the total diameter (p<0.0001), r=0.461 for the solid-part diameter, and r=0.307 for the solid-part ratio (p<0.0001). Patients with total GGN diameters ≥20 mm were significantly more likely to have an SUVmax≥2.5 than were patients with smaller lesions (p<0.0001). No pure GGN or lesion with a solid-part diameter <4.55 mm exhibited an SUVmax≥2.5. Eight tumors with an SUVmax<2.5 were classified as pure GGNs (AIS, 3; MIA, 4; and LPA, 2). There was no significant difference in the frequency of SUVmax≥2.5 in the AIS-MIA group and the LPA-PPA group (p=0.198) (Table II). All had an SUVmax≥1.0 and comprised 38.1% of all pure GGNs observed. The AIS-MIA group showed a significantly lower SUVmax than the LPA-PPA group (p=0.0008). The average value of SUVmax in each group was 0.61 (95% confidence interval 0.309-0.919) in the AIS-MIA group and 1.43 (95% confidence interval 1.07-1.789) in the LPA-PPA group (Fig. 2).

Figure 2. Comparison of SUVmax by each tissue type. *T-test. AIS, adenocarcinoma in situ; MIA, minimally invasive adenocarcinoma; LPA, lepidic predominant adenocarcinoma; PPA, papillary predominant adenocarcinoma.

Table I. Patient characteristics (n=66).

Table I.

Patient characteristics (n=66).

Features	No.
Median age, year (range)	69 (47-86)
Sex (female/male)	38/28
Histology (AIS/MIA/LPA/PPA)	17/15/32/2
pT stage (1a/1b/1c/2a/2b)	30/5/21/9/1
Mean total diameter, mm (range)	19.43 (7.0-41.13)
Mean solid-part diameter, mm (range)	4.55 (0-23.23)
Mean SUVmax, (range)	1.0 (insignificant – 3.9)

[i] AAH, atypical adenomatous hyperplasia, AIS, adenocarcinoma in situ; MIA, minimally invasive adenocarcinoma; LPA, lepidic predominant adenocarcinoma; PPA, papillary predominant adenocarcinoma.

Table II. SUV-max distributions by adenocarcinoma classification (n=66).

Table II.

SUV-max distributions by adenocarcinoma classification (n=66).

SUVmax	<1.0	1.0-1.4	1.5-1.9	2.0-2.4	≥2.5	≥2.5 (%)a
AIS	10	4	1	1	1	5.8
MIA	10	3	1	1		0
LPA	11	8	5	3	5	15.6
PPA	1		1			0

a There was no significant difference in the frequency of SUV_max ≥2.5 in the AIS-MIA group and the LPA-PPA group (p=0.198). AAH, atypical adenomatous hyperplasia; AIS, adenocarcinoma in situ; MIA, minimally invasive adenocarcinoma; LPA, lepidic predominant adenocarcinoma; PPA, papillary predominant adenocarcinoma.

Discussion

Our investigation observed a moderate correlation between the major-axis diameter and SUVmax of GGNs. The potential of using the SUVmax as a reference value was demonstrated by the fact that it was better correlated with the solid-part diameter than the total diameter. However, it is important to note that the correlation was low for pure GGNs and small-diameter lesions. Moreover, in some cases, imaging findings diverged from pathological findings; for example, one case was classified as AIS by postoperative histopathology despite the lesion having a solid part that had been observed before surgery. We did not use high-resolution CT (HR-CT) scanners, but the resolution of our CT images was typical for clinical scanning procedures. Nonetheless, perhaps this divergence would have been less had our scans been performed with greater precision using such equipment.

In this study, we also examined the relationship between AIS, MIA, LPA, PPA and SUVmax. There was a significant difference in the mean value of SUVmax in the AIS-MIA group and the LPA-PPA group, but the clear cut-off value is still unknown. Although the possibility of SUVmax 1.0 being a point dividing the two groups was shown in this study, and care should be taken for GGNs with SUVmax higher than that, further investigation is necessary in the future.

Normal lung tissue has an SUVmax of 0.6 (range, 0.2-1.8) and many past studies have used the criterion of an SUVmax>2.5 to diagnose a finding as malignant when performing PET/CT examinations of pulmonary nodules (11). Since we did not analyze noncancerous lesions in this study, the sensitivity in detecting cancer of SUVmax≥2.5 is unknown. However, in this study, even in the LAP-PPA group, the mean value of SUVmax is <2.5; thus, at least for GGN lesions, it is not an indicator for a non-carcinoma. The SUV is a semi-quantitative value that is affected by factors such as body type, blood sugar levels, and scanning time, and can further vary between different devices or image reconstruction software packages. Because of its simplicity and intuitiveness, it is widely used in clinical procedures to complement visual assessments such as the maximum intensity projection. Even though we analyzed PET/CT scans that were performed at two hospitals, we decided that since the SUVmax was measured at both locations, it could be used as a major assessment measure with a certain level of objectivity. However, we cannot deny the possibility that using scanning data from multiple institutions could have influenced our assessment. The SUV tends to be low in cells with low glucose metabolism on PET images, for example, highly differentiated lung adenocarcinomas and hepatocellular, renal, prostate, and gastric cancers. Past investigations have found that pure GGNs exhibit median SUV values of around 1.0 (4,12); our SUVmax data were not greatly divergent from these values. In addition, PET has low spatial resolution (7-8 mm), making it difficult to assess nodules less than 10 mm in diameter, generally speaking (2,13,14). Moreover, one study found that the SUVmax of lung carcinomas showing localized ground-glass opacities was not significantly different from the SUVmax observed from inflammatory lesions (5). Therefore, we advise caution when using the SUVmax to determine the benignity or malignancy of lung cancers.

PET/CT-based research is becoming more diverse and is utilized in conjunction with a variety of different objectives and use cases. For example, one recent study found that the preoperative SUV could serve as a prognostic factor in c-stage IA lung adenocarcinomas (15). Technological advances are helping to increase the diagnostic accuracy of PET/CT; recent efforts have focused on developing image reconstruction techniques, improving spatial resolution by means of time-of-flight methods, and reducing partial-volume effects (16,17). While the clinical value of PET/CT in evaluating GGNs is still limited at present, it could increase in the future, and thus its evolution must be monitored carefully going forward.

This study is novel in that SUVmax was analyzed based on currently used pathological classifications of lung adenocarcinoma, suggesting the possibility of SUVmax value in lesions with higher malignancy, even in GGN lesions. In order to scientifically confirm the relationship between pathological findings and SUVmax values, future larger-scale prospective observational research with a focus on similar imaging conditions is necessary.

This study was presented at the 2015 World Conference on Lung Cancer (Denver, CO, USA).

Acknowledgements

Not applicable.

Funding

No funding was received.

Availability of data and materials

The datasets generated during and/or analyzed during the current study are not publicly available to protect the patients' personal information but are available from the corresponding author for reasonable requests.

Authors' contributions

KN, AB, NF, YO, SH, MS and MK conceived of and designed the research. KN and AB analyzed and interpreted the data. KN and AB wrote and revised the manuscript. All authors read and approved the final manuscript.

Ethics approval and consent to participate

This study was conducted with the approval of the Ethics Committee of Japanese Red Cross Okayama Hospital. The requirement for informed consent was waived by the ethics committee because of the study's retrospective nature; however, patients could opt out of sharing their information.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Glossary

Abbreviations

Abbreviations:

References