Immune infiltration may predict survival and have clinical significance in lung cancer. However, immune signatures derived from immune profiling based on bulk tumor transcriptomes have not been systematically established in lung adenocarcinoma. We aimed to construct an immune cell infiltrating score, using a new algorithm for evaluating immune infiltration, to improve the prognostic model of lung adenocarcinoma. Public datasets of lung adenocarcinoma from the Gene Expression Omnibus and The Cancer Genome Atlas were adopted as the training and validation cohorts. Fractions of different immune cell subtypes in each sample were estimated using the CIBERSORT algorithm. The immune infiltrating score was further developed by a least absolute shrinkage and selection operator regression model. The prognostic value and clinical relationship of the model was then further explored. An immune infiltrating score model was established on the basis of the immune cells in the training cohort. A high score was associated with significantly worse survival in patients with lung adenocarcinoma (P < 0.001). The prognostic value of the score was confirmed in the validation cohort. The immune infiltrating score could improve the accuracy of predictions of survival when combined with the staging system. Furthermore, the score was potentially associated with patient smoking status and histologic subtype of lung adenocarcinoma. Its possible association with the efficacy of adjuvant chemotherapy was not statistically significant. The immune cell infiltrating score has prognostic significance in predicting overall survival in patients with lung adenocarcinoma.

It is essential to dissect an adequate number of lymph nodes (LNs) to ensure staging accuracy during esophagectomy with or without neoadjuvant therapy. We developed a statistical model to quantify the probability of precise nodal staging based on previous studies. Esophageal cancer patients who underwent esophagectomy were retrospectively reviewed in the Surveillance, Epidemiology, and End Results database. A β-binomial distribution was adopted to estimate the number of understaged patients based on the numbers of positive and examined LNs. Using 6,252 patients, we estimated a 90% confidence of accurate N0 staging could be achieved by examining 17 LNs without neoadjuvant therapy. To obtain similar accuracy in N1 and N2, 20 and 25 LNs should be examined. For patients with neoadjuvant therapy, 18, 19, and 28 LNs could achieve the same accuracy. Staging accuracy was a significant prognostic factor. We found when 90% confidence had been achieved, patient survival did not improve with more LNs examined and the ratio and log odds of positive LNs did not have significant prognostic values. The statistical model we developed for precise staging in patients with different N stages is of great value in guiding lymphadenectomy. It provided risk assessment for underestimated LN metastases and guided subsequent adjuvant treatment.