Reply to Deng et al. and Zhang et al.

We appreciate the interest of Deng et al. [1] and Zhang et al. [2] in our article, in which we demonstrated the importance of level 4 lymph nodes dissection (L4 LND) in patients with left-side non-small-cell lung cancer (NSCLC) [3]. Both letters raised an interesting question regarding L4 LND: Should L4 LND be omitted in some selected NSCLC patients at low risk of mediastinal lymph node metastasis (MLNM)?

In our study, in patients with left lower lobe tumours, the difference in 5-year overall survival (OS) between those with and without L4 LND was not statistically significant. However, the fact that the 5-year OS of patients with L4 LND was 7.8% higher than those without L4 LND needs to be noted (75.1% vs 67.3%, P = 0.16). Because the case number of the left lower lobe group was relatively small, it is reasonable to propose that a statistically significant P-value may be achieved if the case number was large enough in this unselected series. However, we believe that studies based on different tumour sizes and tumour location are necessary to verify if selective L4 LND is acceptable in selected patients.

In a previous study, we demonstrated the incidence and distribution of lobe-specific MLNM in NSCLC using a database composed of 4511 resected cases [4]. The data showed that the incidence of L4 node metastasis was 12.0% in left upper lobe tumours and 11.2% in left lower lobe tumours in unselected operable cases. In our further study, we demonstrated the incidence and distribution of MLNM and its impact on OS in patients with NSCLC 3 cm or less in size, using a database composed of 2292 cases [5]. When tumours were 1 cm or less in size, the incidence of lower zone MLNM for upper lobe tumours and the incidence of upper zone MLNM for lower lobe tumours were zero, which suggested that L4 LND might be omitted in these patients [5]. In this recently published paper [3], however, only 16 patients with tumours at 1 cm or less underwent L4 LND and the small case number did not qualify an OS comparison between the patients who underwent L4 LND and those who did not, just as we discussed as a limitation of the study [3]. We believe that more eligible cases are necessary in future research to enable subgroup analysis based on different tumour sizes.

Based on the previous studies including ours [3–6], we agree that selective mediastinal lymph node dissection (MLND) could be reasonable in selected cases. Nevertheless, the specific standard of selective MLND has not yet been well elucidated. We believe that well-designed multicentre randomized controlled trials are necessary to define the principle of selective MLND. In 2017, a randomized controlled trial comparing lobe-specific MLND with systematic MLND for early-stage NSCLC was commenced (JCOG 1413), which is scheduled to include a total of 1700 patients from 44 Japanese institutions within 5 years [7]. The results of this trial will lend much more data on the efficacy of lobe-specific MLND. At the current time, L4 LND might be omitted in NSCLC patients whose tumours are 1 cm or less in size. But for other indications of selective MLND, we need more data, especially the outcome of JCOG 1413 to elucidate this matter.

REFERENCES