Usefulness of circumference difference for estimating the likelihood of malignancy in small solitary pulmonary nodules on CT

Summary

Objective

The presence of a small solitary pulmonary nodule (SSPN) is a common finding on chest computed tomography (CT); however, preoperative diagnosis of SSPN is often difficult. We measured the extent of ground-glass opacity (GGO) and our own original method of circumference difference (CD) as an additional approach in classifying SSPN revealed on CT, and evaluated the likelihood of malignancy.

Method

In total, 214 single SSPN with diameter <15 mm were studied. All SSPN were histologically examined with surgical resection; preoperative CT findings and pathological diagnosis were evaluated retrospectively. The extent of the ratio of GGO and CD was evaluated using NIH image, where CD is defined as the ratio of the nodule margin distance to the circumference of the predicted circle with an area equal to that of the nodule.

Results

The thresholds for differentiating SSPN are 70% of GGO and 68% of the CD ratio. Differentiation of malignancy from benign tumor using our algorithm had sensitivity of 96.6%, specificity of 86.1%, and positive predictive value of 94.1%.

Conclusion

Combined with other malignant likelihood parameters such as extent of GGO, CD ratio is a useful additional factor for estimating the likelihood of malignancy of SSPN on CT.

Introduction

There is widespread clinical use of computed tomography (CT), which is considered to be one of the most important non-invasive diagnostic techniques, and it commonly detects small solitary pulmonary nodules (SSPN). Evaluation of patients with SSPN, however, is a common and difficult diagnostic problem in daily radiological practice because SSPN may be the first sign of primary lung cancer, especially in its early stage, and the treatment plan differs according to the possibility of malignancy determined on the basis of radiological findings [1]. The prognosis of clinical stage IA lung cancer has a 5-year survival of 67–83% [2] compared with malignant SSPN that has a diameter of 15 mm or less, and thus represents potentially curable disease in which the 5-year survival is inversely related to nodule size at presentation [3].

Large-scale radiographic surveys report that the incidence of malignant pulmonary nodules, as diagnosed by histological examination of the resected nodules, ranged from 30 to 40% before the advent of CT [4]; however, following the use of CT for preoperative diagnosis of pulmonary nodules, the incidence of malignant pulmonary nodules has increased substantially, ranging from 60 to 80% [5], [6], [7]. In our previous study of 120-patients, the incidence of malignant SSPN diagnosed by histological examination following surgical resection was 74.2% [8]. Malignant diagnoses are usually established using invasive procedures such as surgical resection by video-assisted thoracoscopic surgery (VATS) or needle biopsy procedures. Even minimally invasive procedures carry a potentially significant risk of complications and some may also be unnecessary because a population of benign nodules is suspected to exist in the general population; therefore, various measures of the size, shape, CT attenuation, and texture of pulmonary nodules have been utilized by many researchers to distinguish benign from malignant nodules [6], [9], [10], [11], [12], [13]. However, differentiation of benign from malignant SSPN remains a difficult task because many disease processes, ranging from benign growths to primary malignancies, have been associated with the appearance of these generally small, circular objects. The overlapping spectrum of radiographic appearances on CT and clinical parameters between benign and malignant nodules makes differential diagnosis of SSPN difficult in many cases.

Although solid nodules are the most common manifestation of lung cancer on CT finding, nodules with localized ground-glass opacity (GGO) represent early or precursor to adenocarcinoma [14], [15]. Localized bronchiolo-alveolar carcinoma (LBAC) is classified histologically into two groups according to the presence or absence of active fibroblastic proliferation (AFP); LBAC patients without AFP have a better prognosis than those having LBAC with AFP [16]. The extent of the GGO has been researched by several studies that suggest that the prognosis is correlated with the ratio of GGO components because of the radiological–pathological correlation [14], [15]; however, the threshold of the GGO ratio between the two groups (used to speculate on the presence or absence of AFP histologically) has not been clear because previous studies have assumed 50% of the GGO of the two groups to be borderline, based on clinical experience.

In the present study, we measured the extent of the GGO and the circumference difference (CD), which is our own original method, as an additional approach in assisting with classification of SSPN (diameter <15 mm) revealed on CT. We combined the CD ratio results with previously established parameters of the likelihood of malignancy, such as the GGO ratio, and evaluated the usefulness of the CD ratio in predicting the likelihood of malignancy in SSPN.