Comparison of the accuracy of 99mTc-3P4-RGD2 SPECT and CT in diagnosing solitary pulmonary nodules.

The aim of the present study was to compare technetium-99m-(polyethylene glycol-4)3-(Arg-Gly-Asp)2 (99mTc-3P4-RGD2) single-photon emission computed tomography (SPECT) and computed tomography (CT) in the noninvasive differentiation of solitary pulmonary nodules (SPNs). The present study prospectively investigated a consecutive series of 24 patients with SPN, who were newly diagnosed using radiography between September 2012 and January 2014. All patients underwent 99mTc-3P4-RGD2 SPECT and CT scans using a dual-head variable-angle γ-camera equipped with high-resolution collimators. A blinded panel of two thoracic radiologists for CT and three nuclear physicians for SPECT analyzed the images using a 5-grade scale. The SPECT images were also semi-quantitatively evaluated using tumor to non-tumor localization ratios (T/NT). The results were verified by pathological examination of the biopsy material obtained from each patient with SPN, and receiver operating characteristic (ROC) curves were generated from these results. The present results revealed that there were 17 malignant and 7 benign SPNs among the 24 patients with SPN. The mean size of the SPN was 2.1±0.6 cm. Sensitivity of visual analysis for SPECT and CT were 100.0 and 82.4%, respectively, and specificity was 71.4% for the two methods. When the T/NT SPECT semiquantitative analysis (ratio, 1.64) was used as a cut-off, the sensitivity and specificity of SPECT were 100.0 and 71.4%, respectively. The areas under the ROC curves were 0.840 for visual analysis of SPECT [95% confidence interval (CI), 0.600-1.000], 0.849 for semiquantitative analysis of SPECT (95% CI, 0.618-1.000) and 0.815 for CT (95% CI, 0.626-1.000). In conclusion, the present results suggest that 99mTc-3P4-RGD2 SPECT is more accurate than CT in the detection of malignant SPN, and visual analysis appears to be sufficient for the characterization of SPN.

Introduction

Lung cancer continues to be a major public health concern in the USA and other Western countries (1). In 2015, the American Cancer Society reported that 221,200 novel cases of lung cancer were diagnosed in the USA, accounting for ~13% of all cancer diagnoses, and 27% of all cancer-associated mortalities in the USA (2).

A strategy for the reduction of lung cancer-associated mortality rates is to diagnose the disease while it is at an early stage (3). A solitary pulmonary nodule (SPN) is defined as a single lesion in the lung completely surrounded by lung parenchyma with a diameter <3 cm, which is observed on chest radiography and is often the first identifiable manifestation of lung carcinoma (4,5). Previous studies have demonstrated that 68–75% of SPNs are malignant (6,7). Timely and efficient assessment of these nodules is critical for improved patient management (8).

In the USA, >150,000 novel SPNs are identified each year using conventional chest radiography (9). Numerous pulmonary nodules are detected annually using chest computed tomography (CT), since this technique is becoming more widely used in USA patient populations (10). However, CT has limited diagnostic accuracy, since the interpretation of the images relies principally on the size of the lesion and other non-specific findings (11). Magnetic resonance imaging is not a routine examination for the diagnosis of SPN, due to known artefacts that result from tissue-air interfaces and relatively low spatial resolution (12,13). Contrast-enhanced ultrasonography may be used to diagnose pulmonary nodules, but it does not provide a complete image of the lungs due to a variety of reasons, including the lungs being full of gas and interference from the osseous thorax (14). Thus, traditional anatomic imaging does not characterize if the SPN is malignant or benign, which leads to numerous unnecessary surgical lung biopsies (15). Therefore, novel imaging modalities are required to reduce the number of excisional lung biopsies for the diagnosis of SPN.

Fluorine-18-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) is considered to be the most effective method in the differential diagnosis of indeterminate SPN, but its use is limited due to the high cost of the equipment and lack of availability, particularly in developing countries (16). By contrast, single-photon emission computed tomography (SPECT) is more widely used, since it has a lower cost (17). Radioactive tracers for SPECT, including technetium-99m (99mTc)-octreotide acetate (18), 99mTc-depreotide, 99mTc-methoxyisobutylisonitrile (19) and 99mTc-tetrofosmin (20), have been employed in the diagnosis of SPN using diagnostic imaging procedures.

Previous studies have been performed using 99mTc-(polyethylene glycol-4)3-(Arg-Gly-Asp)2 (99mTc-3P4-RGD2) as a novel radioactive molecular probe (21–23). This radiopharmaceutical demonstrated a high binding affinity to integrin αvβ3 in vitro, and exhibited significantly increased tumor uptake and improved in vivo kinetics in animal models (24). In addition, 99mTc-3P4-RGD2 is easily prepared using freeze-dried kits, has a high-labeling yield and radiochemical purity, and does not exhibit adverse events in vivo in nonhuman primates (24). Previously, 99mTc-3P4-RGD2 has been used for the noninvasive differentiation of breast neoplasms (21), and the tracer has demonstrated an impressive image quality with a high sensitivity (83%) in detecting breast cancer.

The primary aim of the present study was to evaluate the use of 99mTc-3P4-RGD2 SPECT in patients with SPN. An additional aim was to compare the diagnostic accuracy of visual and semiquantitative indices of SPECT and CT for the noninvasive differentiation of benign versus malignant SPN using receiver operating characteristic (ROC) analysis.

Materials and methods

Patients

The present prospective study consisted of a consecutive series of 24 patients [14 men and 10 women; age range, 18–77 years; mean age ± standard deviation (SD), 48.79±15.53 years], who had a SPN diameter between 1 and 3 cm on a CT scan that was performed between September 2012 and January 2014 at the China-Japan Union Hospital, Jilin University (Changchun, China). The diameter of all SPNs were calculated based on the maximum lesion diameter observed on CT scanning performed prior to the enrollment of the patients in the present study, which was assessed by thoracic surgeons and radiologists (Department of Radiology, China-Japan Union Hospital, Jilin University). A definitive diagnosis was achieved using transthoracic fine-needle aspiration biopsy or bronchoscopic biopsy. Patients with a malignant SPN underwent surgical resection. Permission to use a novel radiopharmaceutical was obtained from local independent ethics committees and the Institutional Review Board of China-Japan Union Hospital, Jilin University (Changchun, China). Written informed consent to participate in the present scintimammography study was obtained from all patients.

Radiopharmaceutical

The 99mTc-3P4-RGD2 conjugate was generously provided by the Medical Isotopes Research Center of Peking University (Beijing, China) as a freeze-dried kit. 99mTc-3P4-RGD2 was labeled with Na99mTcO4 (China Institute of Atomic Energy, Beijing, China) solution, followed by 20-min incubation at 100°C. Quality control was conducted using radioactive thin-layer chromatography (using an AR-2000 radio-TLC Imaging Scanner, Bioscan, Inc., Washington, DC, USA), which enabled to measure a high labeling yield of ~95%. A molybdenum-technetium generator was provided by Beijing Atom Hi-Tech Co., Ltd. (Beijing, China).

Acquisition of lung SPECT/CT images with 99mTc-3P4-RGD2

99mTc-3P4-RGD2, with a mean radioactivity of 378±86 MBq, was administered via a single intravenous bolus injection in the contralateral arm to the affected lung in each patient, followed by a 10 ml saline flush (Sichuan Kelun Pharmaceutical Co., Ltd., Sichuan, China). Acquisition of the images was obtained at 60 min post-injection. During imaging, patients were in supine position with raised arms. The SPECT/CT system (Precedence SPECT/CT System; Philips Healthcare, Andover, MA, USA) consisted of a dual-head variable-angle γ-camera equipped with low-energy high-resolution collimators and a multi-slice spiral CT component optimized for rapid rotation. Local X-ray scanning (Precedence SPECT/CT System; Philips Healthcare) was performed to image the region of SPN. The following CT scan was set at a matrix of 256×256 pixels, 130 kV, 17 mAs, B60s kernel and a 5-mm slice thickness. Following the collection of CT images, the detecting bench was positioned automatically for SPECT data collection. The matrix was 128×128 pixels and the photopeak was centered at 140 keV, with a symmetrical 20% window. Imaging was performed using 6° angular steps in a 20-sec time frame. The distance between the chest and the detector was minimized as much as possible. Digital Imaging and Communications in Medicine image files of each patient were saved on optic disks (Philips Healthcare) and transferred to Extended Brilliance Workspace 4.5 (Philips Healthcare) for centralized reconstruction, reading and analysis. A fusion SPECT and CT image was produced using Syntegra software (Philips Medical Systems B.V., Eindhoven, The Netherlands).

Evaluation criteria

Three experienced nuclear physicians (Department of Nuclear Medicine, China-Japan Union Hospital, Jilin University) who were blinded to the patients clinical information, physical examinations and radiological findings interpreted the 99mTc-3P4-RGD2 scans individually. The following qualitative interpretation grades were used: Grade 1, no uptake; grade 2, uptake lower than mediastinum; grade 3, uptake equivalent to mediastinum; grade 4, uptake between mediastinum and liver; and grade 5, uptake higher than liver. For semiquantitative analysis, a region of interest was drawn around the entire nodule to the contralateral normal lung tissue. Tumor to non-tumor localization ratios (T/NT) were determined. In the case of grade 1, the T/NT was considered to be 1, due to the absence of uptake. The results were reached by a consensus of the three nuclear physicians.

The CT images were evaluated by two experienced thoracic radiologists who were blinded to the clinical data of the patients. In visual analysis, the image quality of the SPN, including size, shape, conspicuity, margin and presence of calcification, were evaluated. Five diagnostic groups were defined to describe the possible malignancy of each nodule: 1, definitely benign; 2, possibly benign; 3, indeterminate; 4, possibly malignant; and 5, definitely malignant.

Statistical analysis

All numerical results were reported as the mean ± SD. The statistical differences of the T/NT values between patients with malignant and benign nodules were assessed using Student's t test. T/NT values were also compared across the various histological types of lung carcinoma. MedCalc software 12.7.7 (MedCalc Software bvba, Ostend, Belgium) was used to determine the optimal visual interpretation grade of SPECT and CT and the cut-off values of semiquantitative indices of SPECT for the detection of SPN. Linear regression analysis was performed to determine the powerful variables for the prediction of SPN. The incremental diagnostic value of semiquantitative indice analysis was performed using the area under the curve in ROC analysis. P<0.05 was considered to indicate a statistically significant difference.

Results

Patients characteristics

Samples for histological examination were obtained using transthoracic fine-needle aspiration biopsy in 18 patients and bronchoscopy biopsy in 6 patients. In total, 17 out of 24 (71%) SPNs were malignant and 7 (29%) were benign. Among the malignant etiologies, 7 patients had squamous cell carcinoma, 6 adenocarcinoma and 4 small cell carcinoma. The benign aetiologies consisted of 4 tuberculomas, 1 inflammatory pseudotumor, 1 hamartoma and 1 reactive hyperplasia of lymph node. The characteristics of the 24 patients are detailed in Table I.

Table I.

Characteristics of 24 patients with malignant and benign solitary pulmonary nodules.

Patient no.	Pathology	Gender	Age, years	Nodule size, cm	T/NT	SPECT grade	CT grade
1	Squamous cell carcinoma	Female	38	1.6	2.27	4	3
2	Squamous cell carcinoma	Female	27	3.0	3.28	5	5
3	Squamous cell carcinoma	Male	18	2.5	1.87	3	2
4	Squamous cell carcinoma	Female	77	1.9	2.64	4	2
5	Squamous cell carcinoma	Female	39	2.0	2.56	3	3
6	Squamous cell carcinoma	Female	48	1.7	3.05	4	4
7	Squamous cell carcinoma	Male	43	2.8	2.18	3	5
8	Small cell carcinoma	Male	58	2.9	2.95	3	4
9	Small cell carcinoma	Male	48	2.3	2.93	5	4
10	Small cell carcinoma	Male	67	2.2	2.18	3	5
11	Small cell carcinoma	Male	48	2.1	2.22	4	2
12	Adenocarcinoma	Female	57	2.9	2.07	3	4
13	Adenocarcinoma	Female	54	3.0	2.41	4	3
14	Adenocarcinoma	Male	32	1.1	3.27	5	4
15	Adenocarcinoma	Male	52	1.9	2.87	4	5
16	Adenocarcinoma	Male	68	2.7	2.38	4	3
17	Adenocarcinoma	Male	54	1.4	2.77	3	3
18	Tuberculoma	Male	37	1.4	1.04	1	2
19	Tuberculoma	Male	26	2.2	1.41	3	2
20	Tuberculoma	Female	51	2.1	2.18	1	3
21	Tuberculoma	Female	52	2.4	1.10	1	2
22	Inflammatory pseudotumor	Female	67	1.1	3.24	5	4
23	Reactive hyperplasia of lymph node	Male	35	2.3	1.19	1	2
24	Hamartoma	Male	75	1.2	1.06	1	1

Solitary pulmonary nodules were detected using transthoracic fine-needle aspiration biopsy or bronchoscopic biopsy. T/NT, tumor to non-tumor localization ratio; SPECT, single-photon emission computed tomography; CT, computed tomography.

T/NT

In all patients with SPN, the mean T/NT value ± SD was 2.20±0.72 (range, 1.04–3.28). In patients with malignant SPNs, the mean T/NT value ± SD was 2.58±0.43 (range, 1.87–3.28), and in patients with benign SPNs it was 1.60±0.83 (range, 1.04–3.24) (P=0.059). The semiquantitative analysis demonstrated a higher T/NT ratio in malignant lesions compared with benign lesions, although this was not statistically significant. Furthermore, the mean nodule size for malignant and benign SPN was 2.24±0.59 cm, and there was no significant correlation between the T/NT value and the size of the nodule (P=0.92). The various histological types of primary lung carcinoma did not affect the SPECT results [T/NT ratios (Table I): Squamous cell carcinoma, 2.55±0.49; adenocarcinoma, 2.63±0.43; and small cell lung carcinoma, 2.57±0.43]. Representative examples of a high focal uptake in carcinomas are presented in Fig. 1.

Figure 1.

True-positive results for 99mTc-3P4-RGD2 SPECT and SPECT/CT scans, in contrast to a false-positive result for CT. (A) CT, (B) SPECT and (C) CT/SPECT fusion images of a 68 year-old man (patient no. 16) with a 2.7-cm adenocarcinoma and focal uptake of 99mTc-3P4-RGD2 at the inferior lobe of the right lung (tumor to non-tumor localization ratio, 2.38). 99mTc-3P4-RGD2, technetium-99m-(polyethylene glycol-4)3-(Arg-Gly-Asp)2; SPECT, single-photon emission computed tomography; CT, computed tomography.

Optimal cut-off values

The optimal cut-off values for CT interpretation, SPECT visual analysis and semiquantitative analysis were visual score = 3, visual score = 2 and T/NT = 1.64, respectively. The sensitivity, specificity, accuracy, positive predictive value and negative predictive value of the three diagnostic analyses are indicated in Table II.

Table II.

Diagnostic test parameters of three diagnostic methods for the characterization of solitary pulmonary nodules.

Parameters	CT interpretation	Visual analysis of SPECT	Semiquantitative analysis of SPECT
Accuracy, % (95% CI)	79.2 (59.5–90.8)	91.7 (74.2–97.7)	91.7 (74.2–97.7)
Sensitivity, % (95% CI)	82.4 (59.0–93.8)	100.0 (81.6–100.0)	100.0 (81.6–100.0)
Specificity, % (95% CI)	71.4 (35.9–91.8)	71.4 (35.9–91.8)	71.4 (35.9–91.8)
PPV, % (95% CI)	87.5 (64.0–96.5)	89.5 (68.6–97.1)	89.5 (68.6–97.1)
NPV, % (95% CI)	62.5 (30.6–86.3)	100.0 (56.6–100.0)	100.0 (56.6–100.0)

CT, computed tomography; SPECT, single-photon computed emission tomography; CI, confidence interval; PPV, positive predictive value; NPV, negative predictive value.

Visual analysis

The CT scan resulted in 2 false-positive findings (1 tuberculoma and 1 inflammatory pseudotumor) and 3 false-negative findings (2 squamous cell carcinoma and 1 small cell carcinoma) (Fig. 1). Although SPECT visual and semiquantitative analyses also yielded 2 false-positive findings (1 tuberculoma in a different patient and 1 inflammatory pseudotumor in patient 22), it resulted in no false-negative findings (Fig. 2).

Figure 2.

True-negative results for 99mTc-3P4-RGD2 SPECT and SPECT/CT scans, in contrast to a false-negative result for CT. (A) CT, (B) SPECT and (C) CT/SPECT fusion images of a 52 year-old woman (patient no. 21) with a 2.4-cm tuberculoma and low uptake of 99mTc-3P4-RGD2 in the tumor at the middle lobe of the right lung (tumor to non-tumor localization ratio, 1.10). 99mTc-3P4-RGD2, technetium-99m-(polyethylene glycol-4)3-(Arg-Gly-Asp)2; SPECT, single-photon emission computed tomography; CT, computed tomography.

ROC analysis

The empirical ROC areas, which estimated overall diagnostic performance, did not differ significantly among the three diagnostic methodologies (Fig. 3): Semiquantitative analysis of SPECT, 0.849 [95% confidence interval (CI), 0.618–1.000]]; visual analysis of SPECT, 0.840 (95% CI, 0.600–1.000]; CT interpretation, 0.815 (95% CI, 0.626–1.000); semiquantitative analysis of SPECT vs. visual analysis of SPECT (P=0.521); semiquantitative analysis of SPECT vs. CT interpretation (P=0.588); and visual analysis of SPECT vs. CT interpretation (P=0.564).

Figure 3.

Comparison between CT, visual and semiquantitative analyses of technetium-99m-(polyethylene glycol-4)3-(Arg-Gly-Asp)2 single-photon emission CT in the differential diagnosis of malignant and benign solitary pulmonary nodules using receiver operating characteristic analysis. CT, computed tomography; SPECT, single-photon emission computed tomography.

Discussion

99mTc-3P4-RGD2 is a well-designed, dimeric RGD peptide that exhibits increased uptake in mouse cancer xenografts using scintigraphy (25). The uptake of the RGD-containing peptide, quantified by in vitro and in vivo studies, has been observed to be proportional to integrin density and tumor size (26–29). In previous studies, the present authors group applied this novel tracer for the noninvasive differentiation of palpable and nonpalpable breast lesions (21), and the tracer demonstrated an impressive image quality with a high sensitivity in detecting breast cancer.

Chest CT, as a conventional imaging method, continues to be important in the evaluation of patients with SPN (30). Radiological findings that suggest SPN malignancy are thickness of the cavity wall and the presence of a speculated or nodular edge, whereas central, laminated or diffuse calcifications are more likely to be associated with a benign etiology (31). However, despite the advances in anatomic and morphological imaging, numerous nodules remain indeterminate, due to the various criteria employed for their characterization as benign or malignant nodules (7). Invasive and expensive diagnostic procedures, including bronchoscopy and surgical exploration, are often undertaken to obtain a specific SPN diagnosis (32). The observation that 60% of removed nodules are benign lesions indicates the requirement for a simple, efficient and noninvasive approach for the differentiation of benign versus malignant nodules (33).

The present study performed a comparison of 99mTc-3P4-RGD2 SPECT and CT for evaluating SPN. The sensitivity of 99mTc-3P4-RGD2 SPECT was superior to that of CT in the present study; however, the difference in performance of the diagnostic methods was not statistically significant. In total, 17.6% of the nodules were characterized as malignant nodules using 99mTc-3P4-RGD2 SPECT. In addition, 5 SPNs were classified as indeterminate using CT, but were correctly diagnosed using 99mTc-3P4-RGD2 SPECT. Visual analysis yielded the same result as semiquantitative analysis for 99mTc-3P4-RGD2 SPECT, indicating that visual analysis may be sufficient for the characterization of SPN.

When grade 2 was used as a cut-off point for visual analysis, SPECT revealed that all malignant SPNs exhibited focal 99mTc-3P4-RGD2 accumulation with varying intensities. However, the negative predictive value of 100% was possibly due to the low number of SPNs characterized (n=7). Thus, an additional study with a larger patient population is required to determine the correct negative predictive value. The sensitivity reported in the present study for 99mTc-3P4-RGD2 SPECT is comparable to that of 18F-FDG PET/CT (83–100%) (34,35). This high sensitivity may be explained by a high prevalence of malignant tumors in the present cohort. Furthermore, the low 99mTc-3P4-RGD2 uptake in the thoracic region also exhibited a high sensitivity for the detection of malignant nodules. Previous studies have recommended targeting 99mTc-3P4-RGD2 to regions of angiogenesis during tumor development, since a tumor size of 0.2–0.3 cm may exhibit angiogenesis (36). Hypothetically, a pharmaceutical with enough binding affinity such as 99mTc-3P4-RGD2 may detect the majority of tumors with a diameter >1 cm, and may also be the cause of high sensitivity.

In the present study, 1 case of tuberculoma and 1 case of inflammatory pseudotumor exhibited focal 99mTc-3P4-RGD2 uptake. Previous studies have demonstrated that integrin αvβ3 is only observed on the luminal surface of endothelial cells during angiogenesis (37); however, angiogenesis is nonspecific in various pathological events (38). By contrast to other benign lesions, inflammation always exhibits a high cell density and vascularity, which are most likely to be responsible for the increased uptake of the tracer (39). Furthermore, integrin αvβ3 exists on neutrophils, monocytes and vascular smooth muscle cells (40,41); therefore, also contributing to the false-positive results using 99mTc-3P4-RGD2 SPECT.

In conclusion, the present results suggest that a 99mTc-3P4-RGD2 SPECT scan is the most beneficial method for the detection of malignant SPN, and visual analysis appears to be sufficient for the characterization of SPN.