This single-center RCT was accepted through our center’s Institutional Review Board. All participants gave written, informed-consent. This RCT was listed within ClinicalTrials.gov (NCT04217655).
From June 2020 to December 2020, consecutive eligible patients with LNs were randomly assigned into low-dose and standard-dose groups. The inclusion criteria consisted of: (a) clinical cases with LNs, detected on CT, (b) solid LNs, (c) LNs > 5 mm, (d) LNs having intermediate-high risk for lung cancer, depending upon clinical/radiology-based characteristics . Enclusion criteria consisted of: (a) patients who underwent CT-guided CNB previously, (b) LNs which were stable in size for at least 1 year, (c) LNs which decreased in size during follow-up, (d) clinical cases having intense cardiac, pulmonary, renal or coagulatory dysfunctional conditions, and (e) patients who refused to join this RCT.
The primary endpoint of this RCT was diagnostic accuracy, while the second endpoints included technical success, diagnostic yield, operative time, radiation dose, and CNB-related complications.
Randomization and blind
The eligible patients were randomly assigned into 1:1 low-dose and standard-dose groups through block-randomization technique (block size: 8), whereby randomized computer-generated numbers were placed within sequentially-numbered, opaque, sealed envelopes. Prior to biopsy, envelopes were opened by a member of the Science and Education department without any further role in the trial. This RCT was single-blind for the patients.
CT-guided CNB protocols
All procedures were performed under the guidance of a 16-row CT (Philips™, Ccategoryand, OH, USA) operated through a CT-guided interventional radiology expert (10+ years). Table 1 showed the scanning parameters of 2 groups.
The patients’ position was decided according to the sites of LNs. The needle-paths were chosen depending upon preoperative CT outcomes. Co-axial technique was employed during procedures. A 17G outer needle (DuoSmart™, Modena, Italy) was pierced within lung-parenchyma, followed by a second CT scan for establishing needle-tip displace it accordingly. Once outer needle-tip touched LN, an 18G inner semi-automatic core-needle (Wego™, Weihai, China) was inserted via outer needle to obtain the samples from the LNs. Each LN was obtained for 3-4 samples, consequently submerged within 10% formaldehyde until pathology assessments were done.
After biopsy, another CT intervention was conducted for assessing the procedure-linked issues.
Two radiologists evaluated imaging standards independently. One radiologist had 15 years of experience in CT-guided intervention and another one had 8 years of experience in CT-guided intervention. Imaging-quality was evaluated across four categories: category A: needle/LN were distinctly observable, category B: needle/LN were adequately observable, category C: needle/LN were only somewhat observable, and category D: needle/LN could not be seen [12, 13]. Category A and B could be used for CT-guided CNB procedures. If category C or D images occurred, tube voltage and/or current were adjusted to obtain the higher quality images. However, the procedures should be considered as technical failure.
Definitions and diagnoses
LN was defined as spherical/oval image of non-transparent lesions ≤ 3cm in diameter with neighboring pulmonary parenchyma/non-linked to atelectasis, mediastinal lymphadenopathy, or pleural effusion . Technical success for CT-guided CNB was confirmed once pathologists placed their diagnosis from extracted specimens . CNB-based diagnoses could be classified into four categories: (a) malignancy, (b) suspected malignancy, (c) specific benignity, and (d) non-specific benignity. Suspected malignancy was defined as atypical cells suspected for indicating malignancy . Specific benignity was defined as dataset outcomes suggested defined benign-diagnosis, including hamartomas and tuberculosis . Non-specific benignity was defined as benign pathology characteristics existed, through did not suffice for a formal diagnosis .
Resection could make the final diagnoses for both malignant and benign LNs. CNB-based malignancy and specific benignity could be accepted as the final diagnosis [8-13]. CNB-based suspected malignancy and non-specific benignity could not be accepted as the final diagnoses, if they were not confirmed by resection, the CT medical observation would be useful for attaining a finalized diagnostic outcome. Whenever a LN with ≥ 20% size-reduction (without anticancer treatments), or maintained dimensions (no change or decreased < 20%) for a 12 month-minimum period (with no anti-cancer treatments), the final benign diagnosis could be accepted [6, 14].
True-positive was postulated when CNB-based malignancy/suspicious confirmed as malignant at finalized-diagnosis, and true-negative was postulated when CNB-based benignities confirmed benignities at finalized-diagnosis. Diagnosis yield = (CNB-based malignancy + CNB-based specific benignity)/all cases. Diagnostic accuracy = (true positive + true negative)/all cases with the final diagnosis. Pneumothorax and lung hemorrhage were assessed by chest CT. Lung hemorrhage was deemed a novel consolidating/ground-glass opacity around needle tract . High-grade hemorrhage is defined as the width of needle tract hemorrhage > 2 cm .
The sample size was calculated based on diagnostic accuracy with the non-inferiority analysis. Based upon past investigations linked to CT-guided CNB for LNs, we estimated that the diagnostic accuracy was 94% [6, 8, 12]. Based on the −10% of non-inferiority margin with the one-sided significance category of 0.025, we estimated that 200 patients (100 patients per group) were needed after considering 10% drop out rate.
Intention-to-treat (ITT) evaluations were performed depending on total patient group quantity enrolled in this study, while per-protocol (PP) evaluations were performed depending total patients who achieved technical success and definite final-diagnoses.
The continuous data were compared through independent sample t test, while categorical data were compared through Pearson χ2/Fisher exact test. Predictive indicators for diagnosis accuracy and complications were found through univariate and multivariate logistic regression tests. Kappa analysis was conducted for assessing inter-observer agreement regarding imaging-quality. All statistical analyses were conducted through SPSS® v.16.0 (SPSS Inc™, Chicago, Illinois, USA).