Glucose-corrected standardized uptake value (SUV_gluc) is the most accurate SUV parameter for evaluation of pulmonary nodules

Abstract

Standardized uptake values (SUVs) of ¹⁸F-fluorodeoxyglucose positron emission tomography (FDG PET) are widely used to help characterize pulmonary nodules. The purpose of this study is to assess the accuracy of the SUV corrected by blood glucose levels (SUV_gluc), compared to four other commonly used semi-quantitative methods: maximal SUV normalized to body weight (SUV_max), ratio of SUV of nodule to cerebellum (SUV_cer), SUV normalized to body surface area (SUV_bsa) and SUV normalized to body mass index (SUV_bmi). 52 patients with lung nodules had FDG PET scans, consecutively imaged between 7/1/2015 and 6/7/2016. Histopathologic result of the nodules, obtained within two months after the FDG PET scan, demonstrated 10 benign and 42 malignant lung nodules. The SUV_gluc was defined as SUV_max × blood glucose level/100. The average SUV_max was 2.8 for benign nodules and 7.7 for malignant nodules. No significant difference in the receiver operating characteristic (ROC) area under the curves (AUCs) were found between the SUV_max (0.84) and the SUV_cer (0.87) or SUV_bsa (0.86), or SUV_bmi (0.86) with p-values greater than 0.05; however, the ROC AUC for the SUV_gluc (0.90) was larger than that for the SUV_max with p-value of 0.03. These results suggest that SUV_gluc may assist in more accurately representing the glucose metabolism of malignant lung nodules by accounting for the patient’s blood glucose level (BGL). The simplicity of the SUV_gluc method avoids an additional reference ROI, uses preexisting clinical data, i.e. pre-injection blood glucose level, and retains the familiar SUV reference values.

Introduction

¹⁸F-fluorodeoxyglucose positron emission tomography (¹⁸F-FDG PET) is a commonly used imaging modality to assess the risk of benign versus malignant pulmonary nodules, noninvasively [1]. Although the accuracy of PET for diagnosing malignancy is heterogeneous, it is widely accepted for the clinical diagnosis and staging of lung cancer in patients with suspicious lung nodules [2]. Standardized uptake value (SUV) is a simple method to obtain semiquantitiative index of FDG uptake, however multiple factors can affect SUV, thus limit its reliability including: body surface area, lean body mass, blood glucose level, or other perturbing factors [3]. In this study, we compared accuracy of SUV_max (normalized by body weight) and four other corrected SUV parameters including: ratio of SUV of nodule to cerebellum (SUV_cer), SUV normalized to body surface area (SUV_bsa), SUV normalized to body mass area (SUV_bmi) and SUV corrected by blood glucose level (SUV_gluc).

Materials and methods

Patient selection

This study was designed as a retrospective single center study in the University of California San Diego medical center including Hillcrest and Thornton hospitals. It was approved by the institutional review board (IRB) and was Health Insurance Portability and Accountability Act (HIPAA) compliant. Documentation in our database was anonymous. Patients were considered eligible for this study if they underwent FDG PET-CT study between July 2015 and June 2016 and had a pathological diagnosis of the nodule within 2 months after the imaging. The various semi quantitative methods were compared to the “gold standard” which was defined as subsequent histological pathology confirmation.

FDG PET imaging

All patients were asked to fast for at least 6 hours prior to their scan. Blood glucose levels were measured immediately before the FDG injection. Patients were intravenously injected with 370-740 MBq FDG, within a 5-10 second interval. Following an uptake period of approximately 1 hour in a quiet room at rest, multi-station 3-dimensional (3D) PET acquisition with CT, for attenuation correction, was performed for approximately 60 min, using a GE Discovery VCT scanner. PET images were acquired, after the CT scan, at a rate of 2 minutes/bed position, in the 3D acquisition mode. CT images were then reconstructed onto a 512 × 512 matrix. PET images were reconstructed using a standard whole body 3D iterative reconstruction: 2 iterations; 28 subsets onto a 128 × 128 matrix with attenuation correction, decay correction, and scatter correction. The photon energy window was 425-650 keV. Slice thickness was 3.27 mm and reconstruction diameter was 70 cm. Pixel size was 5.47 mm × 5.47 mm with spatial resolution of 5 mm.

Image analysis

All PET images were reviewed and further analyzed using the Agfa Impax software by a board certified academic nuclear medicine physician. Focal activity corresponding to the pulmonary nodule on CT was manually identified on PET images. SUV of the dominant nodule was obtained by manually placing a circular ROI at the site of the maximum FDG uptake in the PET images and the maximal activity (SUV_max) was recorded. SUV_max was calculated as decay-corrected activity of tissue volume (kBq/mL)/injected FDG activity per body mass (kBq/g). SUV_bmi was calculated from SUV_max by normalizing activity based on body mass index (BMI) = Weight (kg)/Height² (m). SUV_bsa was calculated from SUV_max by normalizing activity based on body surface area (BSA) = (Weight (kg) * Height (cm)/3600)^1/2. The average SUV value of the cerebellum reference region was used to calculate the SUV_cer, defined as the ratio of SUV_max divided by the average cerebellar SUV. Corrected SUV for the blood glucose level (SUV_gluc) was calculated based on BGL immediately before the FDG injection (Table 1).

Table 1.

Definition and area under the curve (AUC) in the receiver operating characteristic (ROC) curve of SUV parameters

SUV parameter	Definition	AUC in the ROC curve	p-value**
SUVmax	decay-corrected activity of tissue volume/injected activity per body mass	0.84	Not applicable
SUVbmi	(SUVmax/body weight) × BMI	0.86	0.15
SUVbsa	(SUVmax/body weight) × BSA	0.86	0.43
SUVcer	SUVmax/SUVcerebellum	0.87	0.32
SUVgluc	SUVmax × blood glucose level/100	0.90*	0.03***

^*SUV_gluc has the highest AUC among semiquantitative parameters.

^** p-value is in comparison to SUV_max.

^***SUV_gluc is the only SUV parameter which significantly improves diagnostic accuracy of SUV_max in differentiating benign vs. malignant lung nodules (P = 0.03).

Statistical analysis

All data were expressed as mean ± SD (standard deviation). Differences were analyzed by the paired T-test and considered to be significant at a P-value less than 0.05. Since the sensitivity and specificity of a test depends on the selected threshold value, a more rigorous comparison of diagnostic accuracy was performed using ROC analysis using ROCKIT (1.1B2, University of Chicago, IL, USA).

Results

Histopathological and patient characteristics analysis

Following ¹⁸F-FDG PET, the final diagnosis was confirmed pathologically by subsequent biopsy within 2 months in all 52 patients. It revealed 42 malignant nodules, consisting of 20 adenocarcinoma, 15 squamous cell carcinoma, and 7 other malignancies. The other 10 nodules were benign. The average glucose level was 104.8 mg/dL (range 77 to 235). The rest of the patient characteristics are shown in Table 2. Weight, Height, BGL, BMI and BSA were not significantly different between benign and malignant pulmonary nodules (P>0.05). All five SUV parameters were significantly different between benign and malignant nodules (P<0.05) with SUV_gluc having the smallest p value = 0.0005 (Table 2).

Table 2.

Patient characteristics and SUV values. Mean and Standard deviation (parentheses) values are reported

	All patient	Benign	Malignant	p-value
N, number	52	10	42
Weight, kg	72.8±17.6	73.9±18.1	72.6±17.6	0.83
Height, cm	168.3±9.5	164.0±9.5	169.3±9.3	0.12
Blood glucose level, mg/dL	104.8	92.9±8.4	107.7±32.2	0.16
Body Mass Index, kg/m2	25.6±5.4	27.4±6.4	25.2±5.1	0.24
Body Surface Area, m2	1.83±0.25	1.82±0.25	1.84±0.25	0.88
SUVmax	6.8±4.8	2.8±1.7	7.7±4.8	0.003*
SUVbsa	0.18±0.13	0.07±0.04	0.20±0.13	0.003*
SUVcer	1.5±1.2	0.60±0.57	1.8±1.2	0.005*
SUVbmi	2.4±1.7	1.0±0.6	2.7±1.7	0.003*
SUVgluc	6.8±4.5	2.6±1.6	7.9±4.3	0.0005*

Mean ± standard deviation (parentheses) values are reported.

^*All SUV parameters are significantly different between the benign vs. malignant pulmonary nodules, whereas weight, height, BGL, or BMI are not.

Diagnostic value of five SUV parameters in differentiating benign and malignant pulmonary nodules

We found SUV_max to have the lowest diagnostic accuracy for detecting malignant lung nodules with area under the curve (AUC) = 0.84, in the receiver operating characteristic (ROC) curve, compared to other four semiquantitatitve corrected SUV parameters. The use of alternative semiquantitative methods including SUV_bsa (AUC = 0.86), SUV_bmi (AUC = 0.86), and SUV_cer (AUC = 0.87), increased AUC in the ROC however this increase was not statistically significant (P>0.05), however only for SUV_gluc (AUC = 0.88) this increase was statistically significant (P = 0.03) (Figure 1). Therefore AUC for SUV_gluc showed the highest diagnostic accuracy for detecting malignant lung nodules. Figure 3 shows a representative FDG PET projection image of a patient with left upper lobe lung adenocarcinoma, BGL = 166 mg/dL, with SUV_max = 6.3 and SUV_gluc = 10.5.

Figure 1.

The ROC curves of SUV_gluc, SUV_cer, SUV_bmi, SUV_bsa and SUV_max. AUC for SUV_cer, SUV_bmi, and SUV_bsa is not significantly different from SUV_max (P = 0.32, 0.15, and 0.43, respectively). The SUV_gluc has the largest AUC, significantly different from SUV_max (AUC = 0.90 vs. 0.84, P = 0.03), thus the most accurate SUV parameter to distinguish malignant from benign pulmonary nodules.

Figure 3.

FDG PET projection image in a representative patient with left upper lobe lung adenocarcinoma (arrow), BGL = 166 mg/dL, SUV_max = 6.3 and SUV_gluc = 10.5.

Discussion

The radiopharmaceutical tracer FDG follows a three compartment model with a net uptake rate of K = k1k3/(k2+k3) (Figure 2). SUV_max is proportional to this uptake rate (K) whereas glucose metabolism rate (GMR) is proportional to K × [Glucose] [3,4]. The rational for SUV_gluc is analogous to the calculation of the GMR which involves the scaling of the FDG uptake rate with the blood glucose level. Therefore, SUV_gluc defined as SUV_max × BGL (mg/dL)/100 is a better marker of GMR, especially in higher blood glucose levels, where SUV_max is of limited value. Glucose and ¹⁸F-FDG compete to enter the cells using the same glucose transporters. High BGL reduces ¹⁸F-FDG uptake in the tissue by competitive inhibition and by altered the biodistribution of FDG. The reason for fasting prior to the start of the PET study is to achieve low blood glucose levels for better target-to-background image contrast [5-7]. The end effect is the high blood glucose falsely reduces SUV_max, but not SUV_gluc, due to competitive inhibition of FDG uptake. Although prior publications have shown the advantage of SUV_gluc over SUV_max in evaluation of lymphoma patients [8], predicting the prognosis of pancreatic cancer [8], and brain tumors [9], one study found no advantage of SUV_gluc over SUV_max on SUV-survival correlation in esophageal cancer [10]. It is unclear whether glucose normalization improves diagnosis accuracy or treatment response monitoring of malignant tumors [11]. To our knowledge, only one study has reported the application of the SUV_gluc in lung nodules, where the SUV_gluc was found not to be beneficial in differentiating lung nodules; however, only patients with glucose levels less than 150 mg/dl were studied [12]. We speculate that limiting the patients in that study to those with BGL<150 mg/dL masked the statistical significance of the advantage of SUV_gluc over SUV_max because this advantage is related to BGL as SUV_gluc = SUV_max × BGL/100 (Table 1). In fact, patients with normal BGL do not need glucose correction and those with high BGL will benefit from SUV_gluc. Therefore, to overcome this shortcoming, we included all the patients with BGL ranging from 77 to 235 mg/dL. Another study suggests that for BGL<200 mg/dL, correction of blood glucose is not necessary however the accuracy of SUV_gluc versus other SUV indicators for differentiating lung nodules was not evaluated [13]. Limitations of our study include the retrospective nature and having a single center study; however, it helps the uniformity of the data. Also, we did not evaluate the SUV in the same patient after injection of glucose, to raise BGL and study the effect of BGL on the same patient’s pulmonary nodule.

Figure 2.

FDG tracer kinetic model, follows a three compartment model with a net uptake rate of K = k1k3/k2+k3. C_p = plasma concentration of FDG tracer, C_E = extracellular concentration, C_T = Tissue concentration.

In summary, our study demonstrates that the SUV_gluc is the most accurate semiquantitative method to differentiate malignant from benign lung nodules. The method is relatively simple to adopt clinically since the BGL is readily available as part of routine PET protocols, and there is no need for an additional reference ROI to define.

Disclosure of conflict of interest

None.