Clinical significance of part-solid lung cancer in the eighth edition TNM staging system

The tumour, node and metastasis (TNM) staging system has been one of the most practical and important prognostic predictors of lung cancer for several decades [1, 2].


INTRODUCTION
The tumour, node and metastasis (TNM) staging system has been one of the most practical and important prognostic predictors of lung cancer for several decades [1,2]. Clinicians generally decide the therapeutic strategies for their patients with non-small-cell lung cancer (NSCLC) based on the clinical TNM stage. The recent revision of the eighth edition of the TNM staging system was made based on a large clinical database of 94 708 cases that were collected from 16 countries over the world [3]. Among the 3 descriptors of the staging system, the T descriptor was most greatly modified in the latest revision [4]. T1 and T2 were finely divided by 10 mm each for up to 50 mm in tumour diameter, and T3 and T4 were also re-categorized by tumour diameter as well as the extent of invasion. Moreover, the T descriptor assessment of part-solid tumours (PSTs) was dramatically changed in this revision, shifting from whole tumour diameter to solid component (SC) diameter [5]. In this revision, the ground-glass component (GGC) on thin-section computed tomography (CT) scan and the lepidic growth component of cancer cells in pathological specimens have been eliminated from the tumour diameter, based on findings that invasive size alone more precisely predicts patient survival [6,7]. Although the new T descriptor has been practically used worldwide since January 2017, there are still limited data for validating the descriptor. Recently, several studies indicated that PST composed of both SC and GGC had a more indolent character than pure solid tumour (ST) regardless of the GGC size in the tumour [8,9]. In the present study, we reviewed thin-section images of preoperative CT scans of patients with NSCLC in 2 institutions in Japan to investigate the prognostic significance of PSTs in the clinical T (cT) descriptor of the eighth edition of the TNM staging system.

Patients
Consecutive NSCLC patients who underwent lung resection at Oita University Hospital between 2006 and 2014 (n = 828) and Kyushu University hospital between 2003 and 2012 (n = 710) were retrospectively reviewed. Patients who received any preoperative treatment were excluded. The complete clinical data of 1061 NSCLC patients were available for the assessment of thinsection CT images and prognostic analyses. The tumours included 809 adenocarcinomas, 197 squamous cell carcinomas, 31 large cell carcinomas, 18 adenosquamous carcinomas and 6 other types. The surgical procedures consisted of 298 sublobar resections, 746 lobectomies and 17 pneumonectomies. Nine hundred thirty-one patients had tumours that were pathologically proven to have no lymph node metastasis (pN0). Among N0 patients, the tumours included 730 adenocarcinomas, 163 squamous cell carcinomas, 20 large cell carcinomas, 12 adenosquamous carcinomas and 6 other types. The surgical procedures for N0 patients consisted of 292 sublobar resections, 627 lobectomies and 12 pneumonectomies. Histological diagnosis of tumour cell type was based on the World Health Organization histological classification of lung tumours (fourth edition, 2015). Basically, pulmonary lobectomy with hilar and mediastinal lymph node dissection was selected for lung cancer resection except for Tis or T1mi tumours. Sublobar resection was selected for Tis or T1mi tumours or for patients with limited pulmonary function for lobectomy according to a surgeon's preference.

Ethical statement
The institutional review boards of Oita University and Kyushu University approved the protocol of this study (approval no. 1567 and no. 24-173, respectively) with a waiver of informed patient consent.

Evaluation of the computed tomography images
Preoperative thin-section CT scan findings were reviewed by M.M. and T.O. for patients of Oita University Hospital and by G.T. and T.O. for Kyushu University Hospital. If the independent assessments did not agree, it was reviewed together to achieve consensus. Tumour size was determined preoperatively based on these thin-section CT findings. The condition of CT examination and the measuring method were previously described [10]. Briefly, chest CT was performed in the supine position during inspiratory breath-hold using various multidetector row scanners: Aquilion 4, Aquilion 64, Aquilion ONE and Aquilion ONE VISION (Toshiba Medical Systems, Tochigi, Japan); Somatom Plus 4 Volume Zoom (Siemens Medical Solutions, Malvern, PA, USA); and Brilliance CT and Brilliance iCT (Philips, Amsterdam, Netherlands). The imaging parameters for thin-section CT were as follows: peak tube voltage 120 kVp; tube current 100-500 mA; scan field of view 320-360 mm; and slice thickness 1 mm (Oita University Hospital) or 2 mm (Kyushu University Hospital). All of the CT data sets were transferred to a Picture Archiving and Communication System (PACS). The diameter of the pulmonary tumour was measured manually using axial 2-dimensional CT data at a 1-mm slice section (Oita University Hospital) or 2-mm slice section (Kyushu University Hospital). The tumours with a ratio of SC size-to-whole tumour size (STR) = 0, those with 0 < STR < 1 and those with STR = 1 were defined as pure ground-glass tumours (GGTs), PSTs and STs, respectively (Supplementary Material, Fig. S1).

Patient follow-up
Patients underwent physical examination; laboratory examination for blood cell counts, serum chemistry and serum tumour markers including carcinoembryonic antigen; and chest radiography every 3-6 months after the operation. The chest and upper abdominal CT scans were performed once or twice per year. Brain magnetic resonance imaging and bone scintigrams or fluorodeoxyglucose positron-emission tomography scans were performed annually. The median follow-up period was 5.1 years. A total of 196 patients dropped from the follow-up within 3 years after surgery. Two hundred twenty-two events were calculated in the survival analyses.

Statistical analysis
Qualitative variables were compared using v 2 tests and Fisher's exact test. For quantitative variables, one-way analysis of variance and Student's t-test were applied for each group. Overall survival was defined as the time from the initial surgery until death from any cause. Survival curves were obtained using the Kaplan-Meier method. Significant differences among subgroups were compared using the log-rank test. The Cox proportional hazard model was used to explore the effects of the clinicopathological variables and other factors. Analyses with P-value < 0.05 were considered statistically significant. Statistical analysis was performed using JMP version 11.0.0 (SAS Institute Inc., Cary, NC, USA).
Demographic and clinicopathological characteristics of patients with ground-glass tumour, part-solid tumour and solid tumour The clinicopathological characteristics of all patients are described according to tumour type in Table 1. There were significant differences in sex, surgical procedure, histological findings, whole tumour size, SC size and pathological lymph node metastasis among the 3 groups (P < 0.001). Sublobar resection was preferred as the surgical procedure in GGT (67.8%) compared to PST (34.1%) and ST (20.5%). All GGTs and 99% of PSTs were adenocarcinomas which were significantly different from the The difference between the ground-glass tumour group and the solid tumour group was significant. SD: standard deviation.  proportion of STs, 34% of which were non-adenocarcinomas. All cases of GGT and 96% of PST cases had N0 disease, which was significantly higher than the proportion of ST cases (83%) with N0 disease. The distribution of the size of the solid part component in patients with ST and PST is shown in Fig. 2. Ninety-five percent of PSTs had SC diameters of < _30 mm, whereas 67% of STs had an SC diameter of < _30 mm; significantly fewer PSTs than STs had an SC diameter >30 mm (P < 0.001).
Comparison of postoperative survival of groundglass tumour, part-solid tumour and solid tumour patients according to solid component size

Prognostic impact of part-solid tumour on non-small-cell lung cancer
To assess the prognostic impact of PST on NSCLC, we constructed univariable and multivariable Cox regression proportional hazard models. We evaluated the clinicopathological factors among a combination of patients with PST and ST ( Table 2, n = 974). The multivariable analysis showed that age <70 years, female sex, procedures with a lobectomy or more, SC size, pN0 disease and PST were independent predictors of better survival among all PST and ST patients.

DISCUSSION
In the eighth edition of the TNM staging system, the T descriptor was finely divided by 10 mm each for a tumour diameter up to 50 mm, and the tumour size of PST was changed from whole tumour diameter to SC diameter [4,5]. In addition, pure GGT was defined as Tis (T0), and PST with an SC diameter of < _5 mm was defined as T1mi. Validation studies have demonstrated that the new T descriptor effectively stratified postoperative patients according to the prognoses [11,12]. However, Jung et al. [13] did not observe the superiority of the eighth edition T descriptor system over that of the seventh edition in their validation analysis, suggesting the need for further studies.
The result of the present study supports the superiority of the cT descriptor of the new staging system over that of the seventh edition, with respect to finer stratification of postoperative patients by prognosis.
Since most of the studies that investigated the prognostic impact of PST included only stage IA tumours, data of large-sized PST with SC diameter >30 mm have been limited. In the present study, we demonstrated that only 5% of all PSTs contained tumours with SC diameter >30 mm, in contrast to one-third of STs with SC diameter >30 mm (Fig. 2). These data indicate that the incidence of PST with SC diameter >30 mm was significantly smaller than that of ST with SC diameter >30 mm. We speculate that tumours with a lepidic growth may have less aggressive features that do not grow to a large size in lungs. One report, which included only patients with N0 disease, described a similar proportion; only 3.76% (10 out of 281) of PSTs showed an SC diameter of >30 mm [14]. Nonetheless, to the best of our knowledge, this is the first report to describe the distribution of the SC size of PST among all surgical patients with NSCLC.
Many studies have demonstrated that PSTs with predominant GGC (generally STR < _ 0.5) are less invasive than PSTs with predominant SC (STR > 0.5) or STs [15]. Recently, even patients with PSTs with predominant SC were shown to have better survival than those with the same size of ST [8,9]. Hattori et al. [14] showed that, regarding PSTs, the new cT descriptor (SC diameter) failed to predict prognosis after surgery, although it worked well with STs. In their study, all PST cases had such good survival regardless of SC size that they could not be stratified according to SC size. In the present study, we demonstrated that among patients with cT1 tumours, those with PST with any SC diameter had better survival than those with ST. This result supports the finding that the new cT descriptor failed to distinguish between patient survival curves among PSTs with small-sized tumours. Conversely, regarding the relatively large tumour with SC diameter >30-50 mm, PST did not show a better prognosis to ST tumours (Fig. 4D), although the number of PST patients in this group was very small (n = 13).
Prognostic data of large PSTs with SC size >30 mm have been limited. Recently, Suzuki et al. [16] described the clinical features of PSTs with whole tumour diameter exceeding 3.0 cm. This study indicated that PSTs with predominant SC (0.5 < STR < 1) had a worse prognosis than PSTs with predominantly GGC (0 < _ STR < _ 0.5); the 5-year overall survival rates were 88.4% and 98%, respectively. Although that study did not show how many PSTs had an SC diameter >30 mm, the average SC diameter of the PSTs with predominant SC was 34 mm. Therefore, many PSTs with predominant SC should have an SC diameter >30 mm. Taken together with the present study, PST patients with SC diameter >30 mm had a worse prognosis than those with small-sized PSTs. We would speculate that as the SC enlarges it may acquire more gene mutations and lose its indolent nature, which may reflect the poorer survival of T2 tumours.

Limitations
This study has several limitations. First, it was a retrospective analysis making the study prone to several forms of bias. Nonetheless, to minimize the possible influences from various factors in the survival analysis, we performed a multivariable analysis and demonstrated the prognostic significance of PST with regard to overall survival. In the survival analysis, diseasefree survival was not available; which is also important as well as overall survival for patients with early-stage lung cancer who may die from causes unrelated to their lung cancer. Second, a large portion of the surgical patients had incomplete data and were not included in the analysis. The number of patients included in the sub-categorical analyses may have been too small to yield any definitive results. Third, we used several different detectors and different conditions between the 2 hospitals. We did not perform cross validation of the measurements between the 2 hospitals. The use of different measuring procedures of tumour diameter may have influenced the results. We did not have secondary evaluations of CT images by thoracic radiologists. Fourth, detailed pathological information was not available such as subcategories of adenocarcinoma or invasiveness of the tumour, which could not enable us to see the correlation between the clinical and the pathological T descriptor in the study. Fifth, the new staging system was created from an international cohort of patients while this study contained patients from only 2 centres. Regional differences in oncological outcomes may account for some of the shortcomings of the staging system when applied to this patient cohort, especially when comparing small STs and small SCs of PSTs. Accordingly, a well-designed prospective study with a large number of patients is warranted to clarify the prognostic significance of PST in NSCLC.

CONCLUSIONS
In conclusion, in the detailed analysis of PST and ST for the new cT descriptor, we showed that the incidence of T2a disease or higher is significantly rarer in PST than in ST. PST showed significantly better survival than ST among cT1 tumours, suggesting that the presence of a GGC should be considered in a future improvement of the cT descriptor.

SUPPLEMENTARY MATERIAL
Supplementary material is available at ICVTS online.

Funding
This study was supported by a Grant-in-Aid from the Japan Society for the Promotion of Science (# 18K08791 to T.O.).