Reply to Wang et al.

Abstract

In our study, we defined MVI as microscopic tumour invasion only in relatively large- or medium-sized blood vessels identified by elastic fibres visualized by VVG stain. As Dr Wang et al. pointed out, it is difficult to correctly distinguish capillary vessels and lymphatic channels by the combination of only HE and VVG stains. Tumour cell clusters identified in VVG-negative canals might be invasions into capillaries and lymphatic channels, or retraction artefacts caused by tissue processing. We agree with Dr Wang et al. that immunostaining by specific markers is needed to correctly evaluate tumour cell clusters identified in VVG-negative canals. Mohammed et al. [3] compared immunohistochemical detections of blood and lymphatic vascular invasion by CD31, CD34 and D2-40 with HE stain alone in the breast cancer. The presence of blood and lymphatic vascular invasion was missed in 13% of the breast cancer specimens assessed by HE stain. Hashizume et al. [4] reported the utility of D2-40 immunostaining for detecting lymphatic vessel invasion in p-Stage IA NSCLC. In their study, 18 (16%) of the 114 cases without lymphatic vessel invasion (LVI) on conventional HE and Elastica-van Gieson elastic stains were newly diagnosed as having LVI after adding D2-40 immunostaining. However, the immunohistochemical examinations of vascular invasion and lymphatic permeation are still being investigated in the pathological diagnoses of NSCLC. Although our method based on HE and VVG stains may include some diagnostic uncertainty, it is simple, widely used and more accurate than HE alone. In the systematic review and meta-analysis by Dr Wang et al., 20 (38%) of the 52 studies used the combination of HE and elastic stains to diagnose blood vessel invasion, which is the same as our method [5]. The immunohistochemical examination was adopted in only three (6%) studies. The second question raised by Dr Wang et al. was regarding our study population. Both MVI and MLP are pathological factors of the tumour itself, not of lymph node or of metastasis, and therefore should be incorporated as a T-determinant in the TNM classification system. Therefore, we performed the current study by using N0M0 population to eliminate confounding N and M factors. We agree with Dr Wang et al. that MVI predicts poor prognosis even in stages other than T1a-3N0M0. A previous report from our institution also showed that MVI was an independent risk factor for recurrence (hazard ratio 1.866, P< 0.001) in resected pT1-4N0-2M0 NSCLC [6]. There have been many variations in the methods identifying MVI and MLP. Although we used the terms MVI and MLP, there have been many synonyms indicating lymphovascular invasion in the literature. The unified terminology and standardized diagnostic method of lymphovascular invasion should be established. Based on these, the prognostic impacts of MVI and MLP in resected NSCLC should be investigated by a multicentre study in a large cohort.