The status of LNM in ESCC is essential for evaluating the tumor stage and selection of post-operative treatment(15, 16). The classification on tumor stage of AJCC (7th edition) is primarily based on the number of PLN rather than the number of NLN, since the majority of previous studies have demonstrated that the number of PLN is an independent prognostic factor in ESCC and an indicator for the extent of lymphadenectomy(17).
However, recent reports have shown that the number of NLN also plays an important role in the prognosis of patients with ESCC that have undergone radical resection of ESCC(18, 19). The current study also confirmed that the number of NLN removed in lymphadenectomy was closely associated with the outcome of patients, and further investigation revealed the number of NLN inside the thoracic cavity was an independent predictor for prognosis while the number of NLN outside the thoracic cavity was not an independent prognostic factor. This finding was partly consistent with several studies, which reported that the site of PLN is a more important prognostic factor in ESCC than the number of PLN (12, 20, 21).
Based on the results of the this study, it may not be appropriate to evaluate the prognosis of the ESCC patients only based on the number of PLN. This study identified the RatioNLNinside/PLN not the RatioNLNoutside/PLN or RatioNLNtotal/PLN could be used as an indicator for prognosis in ESCC, which implied that the number of NLN inside thoracic cavity was also an important factor in prognosis. The reason for this is that the number of NLN outside the thoracic cavity could be misleading, since it may include the normal and physiological swollen lymph nodes which usually occur in the abdominal cavity. Furthermore, the number of normal but physiological swollen lymph nodes inside the abdominal cavity is usually greater than that inside the thoracic cavity, which lead to the collection by mistake in lymphadenectomy.
According to the results of our previous research and other studies, lymph node sites around the esophagus had a higher probability occurring tumor metastasis in thoracic ESCC than lymph node stations outside the thoracic cavity, regardless of the location of the primary tumor(22–24). This indicated that the thoracic cavity plays an important role in the tumor metastasis in thoracic ESCC. One reason of this may be that the lymphatic network around the esophagus is the first site of tumor metastasis in ESCC, leading to a high probability of tumor metastasis. Another reason may be that the diaphragm muscle has a blocking effect on tumor metastasis via lymphatic vessels.
This study demonstrated that the combination of the number of NLN inside thoracic cavity and the number of PLN was a strong prognostic factor. Furthermore, on basis of the result of this study, the extent of lymphadenectomy with NLN count inside thoracic cavity 8 times higher of than PLN count were not persuaded to proceed. This finding was consistent to the result of two recent published research, which emphasized that redundant LN removed in lymphadenectomy could not bring benefit for patients` survival (25–26).
Although researchers have tried to set up a modified N stage referring to the principle of N stage in stomach cancer which defined the N stage according to the site of PLN, few reliable results have been obtained until now(27). The reason for this might be that the anatomical structure and site of the esophagus and stomach are different, which leaded to different features of LNM between esophageal cancer and stomach cancer. It may be inappropriate to only use the site of PLN to evaluate the N stage in ESCC. A more reasonable approach would be to incorporate the site and number of lymph nodes consisting of PLN and NLN to evaluate the N stage in ESCC in the future.
At present, it is not clear which site of PLN had a more important effect on prognosis than others. The reason for this may be that it is more reliable to use the number of NLN rather than PLN to evaluate the extent of lymphadenectomy in every regional lymph node site in ESCC, which was supported by a research focusing on the comparison on the impact effect in prognosis between the site and the number of PLN(28). Using NLN to evaluate N stage in ESCC was a novel method, but no stable result on the modified N stage only using the number of NLN were reported. The reason might be that the site of NLN should be considered in the modified N stage based on the findings mentioned above. The RatioNLNinside/PLN might be useful parameter in the promotion of the modified N stage in future.
The present study demonstrated that the number of PLN and tumor differentiation were independent prognostic factors, which was consistent with previous results(29). This finding indicated that this study had a low risk of sample selection bias.
The present study revealed that the hazard ratio of tumor differentiation for prognosis was higher than that for tumor T stage or tumor length, indicating that tumor differentiation was more sensitive and stable in the Cox proportional hazard model than tumor length or tumor T stage. The reason for this result might be that the accuracy rate on the tumor length and tumor T stage would be influenced by the detection site of pathology. This finding was consistent with the results of our previous research, which proposed a difference between poorly differentiated ESCC and middle or well differentiated ESCC.
In summary, this study demonstrated that the number of NLN inside the thoracic cavity maybe a stronger indicator for the extent of lymphadenectomy and prognosis in ESCC than the number of NLN outside the thoracic cavity. Furthermore, this study proposed that the RatioNLNinside/PLN basing on the combination of the PLN and NLN was a useful prognostic factor and identified a cut-off number of it, which indicated the appropriate extent of lymphadenectomy for the ESCC patients.
Although this study presented two novel findings, as mentioned above, there were several limitations that should be mentioned. Firstly, since the limited sample size of this study could lead to statistical bias, the research had replicated the result under the strict statistical method avoiding obvious statistical bias. Secondly, this study failed to promote a modified N stage based on the finding of RatioNLNinside/PLN. The reason might have relation with the limit cohort in this study. However, another large sample of prospective investigation has been prepared, for which a novel method on N stage classification system would be achieved in the near future.