Identifying MPN is crucial in the early detection of LC. In this study, we incorporated a 7-AAB panel with the Mayo prediction model in the differential diagnosis of pulmonary nodules and early detection of lung cancer. Our results confirmed the clinical value of this 7-AAB panel in aiding the diagnosis of early-stage lung cancer, as the detection rate was superior to that of traditional tumor biomarkers. We also validated that the 7-AAB panel in combination with the Mayo model significantly increased the sensitivity, but the PPV and specificity could not be improved in comparison with the 7-AAB panel alone or the Mayo model alone in the differential diagnosis of MPN from BPN, whether MPN was at an early or advanced stage. Based on these findings, we suggest that the 7-AAB panel can be used as a biomarker for the early detection of lung cancer and that it can be incorporated with the Mayo model to determine the probability of malignancy of pulmonary nodules.
Novel biomarkers have been discovered and developed for use in early-stage LC screening, such as autoantibody panels, circulating microRNAs—especially small noncoding RNAs (ncRNAs), circulating tumor DNA, DNA methylation, complement fragments, blood protein profiles, or plasma lipid markers from lipidomics.30,31, 32 Among these, the autoantibody panel EarlyCDT-Lung has been reported and validated as an aid for the early detection of lung cancer.18,19 Many studies have investigated the diagnostic value of joint detection with AABs. Our review previously meta-analyzed four studies that measured the EarlyCDT-Lung Test 7-AAB panel (p53, CAGE, NYESO-1, GBU4-5, SOX2, MAGE A4 and Hu-D), showing a sensitivity of 47% (95% CI 0.34–0.60) with a high specificity of 90% (95% CI 0.87–0.93) in the early detection of lung cancer.19,21,33,34,35 In the present study, we investigated a different 7-AAB panel (p53, GAGE7, PGP9.5, CAGE, MAGEA1, SOX2, and GBU4-5), which identified 67.5% of early-stage LC with a specificity of 89.6%. Although these 7-AAB panels possess high specificity as serum diagnostic markers in the diagnosis of early-stage lung cancer, the low sensitivity limits the application of the AAB panels in clinical practice. Thus, there is an urgent need to find approaches that can improve the sensitivity of the detection efficacy of early-stage LC. As noted previously, one recent study13 evaluated the combination of a 7-AAB panel and low-dose computed tomography (CT) scanning and significantly improved the diagnostic yield in early-stage MPN patients, with the PPV significantly improved to 95.0% when compared with the AAB panel alone (95.0% vs. 85.2%; P < 0.001) or with CT scanning alone (95.0% vs. 69.0%; P < 0.001). Another study also found that this 7-AAB panel could distinguish malignant lesions from benign lesions and control cases, with a sensitivity of 56.53% and a specificity of 91.60%, but the specificity could be further increased to 95.80% when combined with CT.22 To overcome the drawbacks of CT’s high false-positive rate and radiologist subjectivity, in the present study, we combined a 7-AAB panel with the Mayo prediction model in the differential diagnosis of pulmonary nodules. The 7-AAB panel showed a sensitivity of 67.5% in the detection of early-stage MPN. However, the 7-AAB panel combined with the Mayo model had a significantly improved detection efficacy when compared with the AAB panel alone or the Mayo model alone, with sensitivities of 93.5% and 91.4% in distinguishing early- and advanced-stage malignant nodules, respectively, from benign nodules, but the PPV and specificity did not improve correspondingly. The results were not consistent with two previous studies that combined this 7-AAB panel with CT scans. 13,22 We hypothesize that this could be related to the high false-positive rate of and subjective diagnosis from CT. For improving sensitivity, we view the 7-AAB panel and the Mayo model as complementary rather than competitive, and the combination of the two methods may be beneficial in distinguishing benign from malignant lesions, particularly early-stage MPN, which is potentially curable when detected early.
Recently, several prediction models including clinical and radiological values have been developed that can help physicians distinguish between benign and malignant nodules.9 A study found that the Mayo, Veterans Association (VA) and Brock models showed similar predictive performance for malignant nodules (AUC: 0.6145, 0.6042 and 0.6820, respectively) and outperform the Herder model (AUC: 0.5567), which includes the 18FDG uptake value.36 Another study evaluated three prediction models, the Mayo, VA, and Peking University (PU) models. The area under the ROC curve of the PU model [0.800; 95% confidence interval (CI): 0.708–0.891] was higher than that of the Mayo model (0.753; 95% CI: 0.650–0.857) and of the VA model (0.728; 95% CI: 0.6230-833); however, these findings were not statistically significant. This means that these mathematical prediction models have similar accuracy for the prediction of SPN malignancy.11 Therefore, we selected the most extensively validated Mayo model to aid in distinguishing between benign and malignant nodules. Moreover, our study applied the Mayo model to separately investigate early- or advanced-stage MPN patients, and the results showed that the model’s AUC was 0.670 (95% CI: 0.605–0.730) for early-stage MPN and 0.933 (95% CI: 0.889–0.964) for advanced-stage MPN, which is in line with the recently reported literature.11, 36 It seems that the Mayo model has greater accuracy for predicting malignant PNs at the advanced stage than at the early stage. We assume that this may be related to the larger nodule size and higher malignancy probability (> 65%) in late-stage MPN. However, when the Mayo model was combined with the 7-AAB panel, the AUCs were significantly improved for both early-stage and advanced-stage MPN (0.795 and 0.950, respectively).
Because traditional TAA markers, such as cytokeratin 19 fragment antigen (CYFRA21-1), neuron-specific enolase (NSE), carcinoembryonic antigen (CEA), and squamous cell carcinoma antigen (SCC), remain widely used as reference diagnostics for lung cancer, 37 we compared the 7-AAB panel with the combination of these traditional TAAs in the diagnosis of LC. We found that the 7-AAB panel was good in the early stages of lung cancer, while the traditional tumor markers showed a higher sensitivity with late-stage LC. This suggests that the 7-AAB panel is not suited for use as a biomarker for late-stage LC patients, whereas the traditional antigen biomarkers such as CEA, NSE, SCC and CYFRA 21 − 1 should not be used to diagnose early-stage LC patients.
Identifying malignant pulmonary nodules and achieving the early detection of lung cancer significantly improves the survival rate and decreases mortality associated with this disease. Currently, the EarlyCDT-Lung test is being evaluated in a large-scale screening study in individuals at high risk of lung cancer worldwide. In addition to validating the clinical efficacy of the 7-AAB panel for the early detection of lung cancer, we also found that the combination of the 7-AAB panel with the Mayo model could significantly improve the sensitivity for distinguishing benign from malignant lesions at both early and late stages. As the ELISA of AABs is relatively low cost, noninvasive, and easy-to-perform, and the Mayo model is defined by equations, a combination based on the 7-AAB panel illustrated here and the Mayo model holds promise for the early detection of MPN, and it can be applied in some undeveloped areas or hospitals without high-resolution CT scans. Early detection of malignancy and timely resection are important while the nodule is still relatively small, as this could lead to decreased mortality.
Inevitably, there are some limitations in our research. First, the number of stage I-II MPN cases may not be sufficient. The BPN group was not matched well with the MPN groups for age, gender or smoking status. Additionally, the study included no other cancer control group aside from lung cancer that could have had some or all AABs in common with those of the panel. Additionally, although the Mayo model was designed for pulmonary nodules, we did not investigate the AAB panel and the Mayo model with different sizes of PNs or subtypes of MPN to further evaluate their validity. Furthermore, we only analyzed the diagnostic efficacy of the AAB panel and the Mayo model in a Chinese population, and future work is ongoing to validate the sensitivity and specificity of the combination in other ethnicities.
In conclusion, our study confirmed the clinical value of the 7-AAB panel for the early detection of lung cancer, which achieved a sensitivity of 67.5% and a specificity of 89.6%. This 7-AAB panel proved to be better than traditional tumor markers, such as CEA, NSE, SCC and CYFRA 21 − 1 in aiding with early diagnosis. The combination of the 7-AAB panel with the Mayo model can improve the sensitivity for distinguishing benign PNs from malignant nodules, but the combination could not improve the PPV or the specificity. Taken together, this study illustrates the robust potential of the 7-AAB panel for the early diagnosis of lung cancer, and in combination with the Mayo model could be used to distinguish the probability of malignancy of pulmonary nodules in clinical practice.