Prognostic significance of ground-glass areas within tumours in non-small-cell lung cancer

Abstract

Open in new tabDownload slide

OBJECTIVES

To validate or refute the hypothesis that non-small-cell lung cancers (NSCLC) with ground-glass areas (GGA+) within the tumour on high-resolution computed tomography are associated with a more favourable prognosis than those without GGA (GGA−).

METHODS

We analysed data from a multicentre observational cohort study in Japan including 5005 patients with completely resected pathological stage I NSCLC, who were excluded from the Japan Clinical Oncology Group (JCOG) 0707 trial on oral adjuvant treatment during the enrolment period. The patients’ medical and pathological records were assessed retrospectively by physicians and re-staged according to the 8th tumour, node, metastasis edition.

RESULTS

Of the 5005 patients, 2388 (48%) were ineligible for the JCOG0707 trial and 2617 (52%) were eligible but were not enrolled. A total of 958 patients (19.1%) died. Patients with GGA+ NSCLC and pathological invasion ≤3 cm showed significantly better overall survival than others. In patients with tumours with an invasive portion ≤4 cm, GGA+ was associated with better survival. The prognoses of patients with GGA+ T2a and GGA− T1c tumours were similar (5-year overall survival: 84.6% vs 83.1%, respectively). The survival with T2b or more tumours appeared unaffected by GGA, and GGA was not prognostic in these larger tumours.

CONCLUSIONS

Patients with GGA+ NSCLC on high-resolution computed tomography and ≤4 cm invasion size may have a better prognosis than patients with solid GGA− tumours of the same T-stage. However, the presence or absence of radiological GGA has little impact on the prognosis of patients with NSCLC with greater (>4 cm) pathological invasion.

INTRODUCTION

The International Association for the Study of Lung Cancer published the 8th edition of the International Union Against Cancer tumour, node, metastasis (TNM) classification of lung cancer in 2017 [1]. One of the major changes in this TNM classification was the pathological categorization of T factor by invasive size, rather than maximum whole-tumour size, excluding portions of lepidic tumour growth [1, 2]. This corresponds to clinical T factor defined by the size of the solid component, i.e. excluding ground-glass areas (GGAs) within the tumour on high-resolution computed tomography (HRCT) [3, 4].

The presence of GGA within the tumour has repeatedly been correlated with a better prognosis in patients with non-small-cell lung cancer (NSCLC) [5–11], and stage IA NSCLC with GGA (GGA+) within the tumour on HRCT had a better prognosis than NSCLC with solid components without GGA (GGA−) [6–9]. Indeed, some researchers have advocated adding information on the presence or absence of GGA within the tumour to the clinical T descriptor in future editions of the TNM classification for lung cancer [8, 12]. However, the correlation between GGA within the tumour and prognosis in patients with larger, T2 and more advanced T-stage lung cancers remains controversial.

In the present study, we carried out a sub-analysis of data from a multicentre observational cohort study (CSPOR LC-03) [13] of patients who were excluded from the Japan Clinical Oncology Group (JCOG) 0707 study [14]. Since the patients were excluded from a clinical trial and treated on daily practice, they reflect the real-world outcome of such patients. As were previously reported [13, 15, 16], CSPOR LC-03 study population is heterogenous, with approximately half are trial-eligible ‘fit’ patients and another half trial-ineligible poor-risk ones, making it suitable to clarify what is actually going in daily practice. The study aimed to elucidate the differences in clinicopathologic characteristics and prognoses in relation to the presence or absence of GGA on HRCT or pathological non-invasive areas (lepidic growth component), based on the 8th TNM classification for resected node-negative NSCLC, using a large cohort of real-world data in patients excluded from clinical trial. Our study purpose was to validate or refute the hypothesis that presence of GGA is, in fact, a favourable prognostic factor, even after matched for pathological invasive tumour size.

PATIENTS AND METHODS

Patient registry

This was an observational multicentre cohort study conducted in Japan. Patients with completely resected pathological stage I (T1 > 2 cm and T2 in 6th TNM edition) NSCLC confirmed by lobectomy or larger resection with mediastinohilar lymph node dissection (i.e. target population of the JCOG0707 trial [14] but excluded from that trial during that study’s enrolment period) were eligible to participate in this study. The eligibility criteria of JCOG0707 is given in Supplementary Material, Table S1 [14]. The enrolment period of the JCOG0707 trial was November 2008 to December 2013. Researchers from institutions participating in the JCOG0707 trial recorded data from the patients’ medical records, as described previously [13, 15, 16]. Of the 48 institutions participating in JCOG0707, 34 participated in the current observational study. The collected data included the following clinicopathologic and prognostic items: sex, age, presence or absence of GGA within the tumour on HRCT (GGA+ and GGA−, respectively), reason for exclusion from JCOG0707 trial, pathological T factor, mode of surgical procedure, mode of lymph node dissection, tumour diameter, histology, pathological invasion size within the tumour, PL (the pathological extent of pleural invasion defined by the TNM classification), survival time, recurrence and cause of death.

All patients were re-staged according to the recent 8th edition of the International Union Against Cancer TNM Classification of Malignant Tumors [1] (published in 2017) regarding T factor, but pathological invasion size was only measured in GGA+ NSCLCs that were >3 cm in whole-tumour size, as GGA+ NSCLCs with whole-tumour size of ≤3 cm should always have a solid-component size of 3 cm or less. Tumour histology was described according to the 4th edition of the World Health Organization (WHO) classification published in 1999, where pathological invasion was defined when the tumour cells were arranged in acinar/papillotubular structures or solid nests in a fibroblastic stroma, often accompanied by collagenization, and when the alveolar structures were no longer discernible. These pathological findings were diagnosed by expert pathologists at each institute. For GGA−, solid tumours, whole-tumour size was considered to be equal to the invasion size, as shown previously [3, 4]. T factor was thus defined based on solid-component size in radiological GGA− tumours, and pathological invasion size in radiological GGA+ tumours with >3 cm whole-tumour size.

GGA was defined as an area of slight, homogenous increase in density that did not obscure the underlying vascular markings on HRCT. Additionally, PL1 tumours not exposed to the visceral pleural surface were not regarded as a T descriptor, even if the tumour invaded beyond the elastic layer of the visceral pleura but did not extend to the pleural surface of the lung, because T category was registered based on the 6th TNM classification in our study. T category was thus assigned based on pathological invasion size, clinical solid-component size or PL2 pathological pleural invasion extending to the visceral pleural surface.

We categorized the tumour type based on the whole-tumour size, the presence or absence of GGA on HRCT and the pathological invasion size (Supplementary Material, Table S2), with types 1–6 considered as the GGA+ group and types 7–11 considered as the GGA− group.

Tumour histology was described according to the 4th edition of the WHO classification (published in 2015) [17]. Mediastinohilar lymph node dissection was performed according to the ‘systematic’ or ‘lobe-specific’ dissection modes [18, 19].

Ethical statement

This study was conducted according to the Declaration of Helsinki and approved by the institutional review board of each participating institute, as well as the ethics committee of the Public Health Research Foundation. This was an observational study, and the need for signed informed consent was waived according to the Japanese Ethical Guidelines on Scientific Research [20]. The trial was registered with the UMIN Clinical Trials Registry (UMIN000015732).

Statistical analysis

Differences in categorical and continuous variables were evaluated by χ² tests and one-way analysis of variance, respectively. Survival time was defined as the time between the date of surgery and death with patients without event being censored at the last follow-up date. Survival curves were estimated by the Kaplan–Meier method, and differences in survival were assessed by the log-rank test. Overall survival (OS) was defined as the time between surgery and death from any cause. We used the reverse Kaplan–Meier survival curve [21], which is constructed by reversing ‘censor’ and ‘event (death)’, to compute the median follow-up period. Multivariable Cox’s proportional hazards analysis was used to adjust the significance of factors essential for prognoses according to clinicopathological background for NSCLC, including sex, age, mode of lymph node dissection, histology (adenocarcinoma versus others), tumour size, p-T status (T1 versus others) and adjuvant therapy (tegafur/uracil and others). Significance was defined as a P value <0.05. All data analyses were performed with SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

Patients

Of the 48 institutions that participated in the JCOG0707 trial, 34 institutions, which enrolled 917 (95%) of the 963 patients registered in the JCOG0707 trial, participated in the current study. During the JCOG0707 trial accrual period, a total of 5922 patients underwent surgical completely resection for pathological stage I (T1 > 2 cm and T2 in 6th TNM edition) NSCLC. Excluding the 917 patients who participated in the JCOG0707 trial, 5005 patients (‘All’ cohort) were enrolled in this observational study. The patient recruitment strategy is shown in Fig. 1. Overall, 85% of patients (5005 of 5922 patients) were excluded from the JCOG0707 trial, including 2388 (48%) who were ineligible for the trial (‘Ineligible’ cohort) and 2617 (52%) who were eligible but were not enrolled for various reasons, including patient refusal or temporary suspension of accrual (‘Eligible’ cohort) [16]. The 5005 patients included 2916 males (58%) and 2089 females (42%), aged 20–93 years (median, 69 years). Approximately one-third of the patients (1667; 33%) received adjuvant therapy. Most of them (1549 or 93% of those with adjuvant therapy) were treated with standard tegafur/uracil. The impact of adjuvant tegafur/uracil on patient outcome has been reported previously [13]. Among the patients who received ‘other’ adjuvant chemotherapy, the majority received platinum-base, and very few (3 patients; 0.1%) received epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitor.

Clinicopathological findings

The clinicopathologic characteristics of the patients in this study and those accrued to the JCOG0707 trial (the ‘JCOG0707’ cohort) have been reported previously [14, 16]. The characteristics of the ‘Eligible’ and ‘Ineligible’ cohorts are re-summarized in Table 1. The ‘All’ cohort comprised 2115 patients (42.3%) of the GGA+ group and 2864 patients (57.2%) of the GGA− group. EGFR mutation status was investigated in 2172 (44%) of the patients, with 831 (17%) mutation positive, and 1341 (27%) wild type. EGFR mutation was observed in 488 (53.5%) of the 906 GGA+ patients, 335 (27.1%) of the 1238 GGA− patients and in 8 patients without information on radiological GGA.

Prognosis

The median follow-up period was 6.0 years. A total of 958 patients (19.1%) died. The 5-year OS rates for the ‘Eligible’, ‘Ineligible’ and ‘JCOG0707’ cohorts were 89.3%, 78.7% and 89.5%, respectively (Supplementary Material, Fig. S1). ‘Ineligible’ patients had significantly poorer OS than patients in the ‘Eligible’ and ‘JCOG0707’ cohorts (P < 0.001), while the survival curves were similar in the latter 2 cohorts, implying that ‘study partition’ itself had no impact on patient outcome.

The OS curves according to the mode of lymph node dissection for ‘All’ patients are shown in Supplementary Material, Fig. S2. The survival curves between patients with systematic and selective lymph node dissection were not overall statistically significantly different with a P-value of 0.644 and survival rates at time 5 years of 84.1% and 84.5%.

The OS curves according to tumour type for the ‘All’ cohort are shown in Fig. 2 and the hazard ratios (HRs) of OS are shown in Table 2. The survival curves of patients with types 5 and 6 tumours (GGA+, and >5 cm invasive portion) and patients with types 10 and 11 solid tumours (GGA− and >5 cm invasive portion) were analysed together as ‘large’ tumours, due to the relatively small numbers of patients.

The 5-year OS rates for the ‘All’ cohort were 94.1% for type 1, 94.9% for type 2, 84.6% for type 3, 73.4% for type 4, 61.6% for types 5 and 6, 83.1% for type 7, 76.8% for type 8, 72.6% for type 9 and 62.3% for types 10 and 11, respectively (Fig. 2, Table 2). Patients with GGA+ lung cancer with invasion size ≤3 cm (types 1 and 2) had significantly better OS than others, with the 5-year OS rate exceeding 94%.

We also analysed the OS according to tumour type for the ‘Eligible’ cohort alone, because the ‘Ineligible’ cohort had many non-lung cancer deaths, which obscures the prognostic impact of the tumours, as previously reported [15].

The 5-year OS rates for the ‘Eligible’ cohort were 96.3% for type 1, 96.0% for type 2, 90.3% for type 3, 91.1% for type 4, 69.6% for types 5 and 6, 88.5% for type 7, 83.6% for type 8, 80.6% for type 9 and 65.3% for types 10 and 11, respectively. These showed the same tendency as those of the ‘All’ cohort.

In the ‘All’ cohort, as shown in Fig. 3, in patients with tumours with an invasive portion ≤4 cm, GAA+ was associated with better survival. The prognosis of GGA+ T2a tumours (type 3) was comparable to that of GGA− T1c tumours (type 7), and significantly better than that of T-stage-matched type 8 GGA− T2a tumours. The 5-year OS rates were: 84.6% for type 3, 83.1% for type 7 and 76.8% for type 8 (type 3 versus type 8: HR 1.5062, P = 0.0443). However, for T2b or larger tumours, there was no significant difference in OS for NSCLCs of the same T factor (types 4 vs 9: HR 0.9594, P = 0.8895, and types 5 and 6 vs 10 and 11: HR 1.0840, P = 0.7786), regardless of the presence (types 4, 5 and 6) or absence (types 9, 10 and 11) of GGA within the tumour (Fig. 4).

As was previously reported [15], OS was significantly better for EGFR-mutant patients by univariable, but not by multivariable analysis (HR 0.834, P = 0.1123), probably reflecting patient backgrounds, such as sex and smoking habits.

DISCUSSION

As is often the case with prospective clinical trials, only a selected population of candidate patients were enrolled in the JCOG0707 adjuvant chemotherapy trial for early-stage NSCLC. On the other hand, those NOT enrolled in it make a ‘real-world’ cohort, consisting of heterogenous patients. To elucidate the similarities and differences of the trial population and non-trial one, we first analysed the patient outcomes in the heterogenous ‘All’ cohort, then validated the findings in more homogenous and likely to be more cancer-specific ‘Eligible’ cohort. The current observational study reproduced the survival outcome of JCOG0707, especially in the ‘Eligible’ cohort. These data thus provide valuable real-world information, both on prospective trial-eligible and ineligible population, on the outcomes and treatment effectiveness in these patients.

The current 8th edition (and the upcoming 9th edition as well) of the TNM classification of lung cancer categorizes clinical and pathological T factors based on measurement of the invasive component alone. As in preoperative radiologic findings, the solid part of the tumour on HRCT is regarded as the invasive component, whereas GGA within the tumour corresponds to non-invasive areas, where the tumour cells grow along the surface of alveolar walls. In the present study, invasion size was measured pathologically in tumours with GGA (GGA group or GGA+), whereas solid areas on HRCT were considered to indicate the invasion size in tumours without GGA (solid group or GGA−). The analytical methods were based on, and consistent with, previous reports [3, 4] showing that clinical T-stage reflected pathological T-stage, especially in GGA− solid tumours.

Although the present TNM classification decides the T factor based on the size of the solid area, regardless of the presence or absence of GGA within the tumour on HRCT, several recent studies [7–10, 22–24] have reported that radiologic solid tumours (GGA− NCSLC) exhibited more malignant behaviour and had a poorer prognosis compared with GGA+ part-solid tumours, especially for T1-stage IA tumours. They accordingly advocated that the oncologic and prognostic outcomes should be discussed separately for lung cancer patients with and without GGA within the tumour. In addition, previous reports showed that the 5-year survival of patients with T1N0M0 (stage IA) lung cancer with GGA on HRCT was similar, regardless of the size of the solid component [8, 11]. The current study also showed that survival differed between T1-stage IA tumours with and without GGA, even if the T factor, defined by invasion size within the tumour, was classified in the same category. GGA− solid T1 tumours (type 7) showed a significantly poorer prognosis than GGA+ T1 tumours, with no survival difference among GGA+ T1 tumours (types 1 and 2) regardless of the whole-tumour size.

Hattori et al. [25] suggested that all tumours with radiological GGA should be classified together as T1a, independent of the whole-tumour size or solid-component size on HRCT, because the prognostic outcome of clinical tumours with GGA was excellent, regardless of these factors. However, their analysis included only a small number of GGA+ tumours with solid portions sized >3 cm.

The current multi-institutional cohort study found that GGA+ tumours with ≤3 cm pathological invasion size had a better prognosis than GGA+ tumours with >3 cm invasion size, so that putting all ‘tumors with GGA’ together would be inappropriate. The prognosis of GGA+ T2a (type 3) tumours was still significantly better than GGA-T2a (type 8) that had the same T factor of type 3, and comparable to that of GGA− T1c (type 7) tumours (5-year OS in ‘All’ cohort: 84.6% vs 83.1%, respectively), as shown in Fig. 3. On the other hand, the prognosis of T2b or larger tumours was unaffected by the absence/presence of GGA. Although the small numbers of patients precluded a definite conclusion regarding the comparison of type 4 versus type 9 tumours in the ‘Eligible’ cohort, the OS of patients with ‘large tumors’ in the ‘All’ cohort was similar to that for NSCLCs with the same T factor, with or without GGA (Fig. 4). The survival impact of GGA within the tumour thus appears to be restricted to tumours with a pathological invasion size ≤4 cm. The prognosis of the tumour with invasive size >4 cm presumably would depend on no longer the presence of GGA but only invasive size (solid component). Further validation studies are necessary to confirm the result in tumours with invasive size >4 cm, especially T2 tumours.

Limitations

This study had several limitations. First, the data were collected retrospectively from multiple centres, and there may have been interobserver differences in terms of the measurement of pathological invasion size or radiological solid size on HRCT among institutions. Regarding GGA+ tumours, GGA within the tumour on HRCT was not confirmed to correspond to a pathologically non-invasive component of the carcinoma. Second, whole-tumour size ≤2 cm NSCLC was not included in the present study, but the clinicopathological features in NSCLC for whole-tumour size ≤2 cm have been elucidated by many prior studies [5–7, 26]. Third, tumour size was regarded as pathological invasion size in GGA− NSCLC, without actual review of histopathological specimens. Fourth, pleural invasion of PL1 was not reflected in the T descriptor because T category was registered based on the 6th TNM classification in our study. Fifth, detailed information on the GGA status of patients in the JCOG0707 cohort was stored in another database and could not be evaluated. In addition, half of the JCOG0707 cohort received investigational adjuvant chemotherapy with tegafur/gimeracil/oteracil (TS1). However, the OS was similar in the JCOG0707 and ‘Eligible’ cohorts and adjuvant TS1 therapy did not affect the prognosis of patients in the JCOG0707 trial [15], suggesting that these factors were unlikely to result in significant bias. Sixth, since no adjuvant tyrosine kinase inhibitor therapy was recommended at the time of the present study period, EGFR mutation status was not thoroughly investigated. Although our study did not show the survival impact of EGFR mutation per se [15], ADAURA study showed potential OS benefit of adjuvant tyrosine kinase inhibitor in patients with EGFR mutation [27], and future studies would have to incorporate the effect of biology-based postoperative therapies. Seventh, truly ‘real-world, all-comer analysis’ might have to be done including those enrolled in the JCOG0707 trial, but due to the different of data management at different data centres, we could not perform such analyses. This could somewhat undermine the generalizability of our data, by excluding JCOG study-participating subset. Therefore, by showing that the current observational study reproduced the survival outcome of JCOG0707, especially in the ‘Eligible’ cohort, we tried to guarantee that our study population represents real-world patients, both prospective trial-eligible and ineligible population. Finally, since our data were not specifically designed to evaluate the correlation with tumour size and patient outcome according to GGA, they should be taken as hypothesis-generating and not conclusive.

CONCLUSION

Patients with GGA+ lung cancer with an invasion size ≤4 cm had a better prognosis than those with GGA− solid tumours, for the same T-stage defined by the 8th edition of the TNM classification. However, the presence or absence of GGA within the tumour had little impact on the prognosis in patients with lung cancer with a pathological invasion size or clinical solid size >4 cm. Our data supports that the presence of GGA within the tumour thus only has prognostic significance in patients with an invasion size ≤4 cm. This warrants further investigations, and if validated, should be reflected in future TNM classifications.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

ACKNOWLEDGEMENTS

We thank all 34 institutions that participated in this multicentre study conducted in Japan. The late Prof Yasuo Ohashi, a Professor of the Department of Integrated Science and Engineering for Sustainable Society in Chuo University, made an invaluable contribution to design and conduct of the study.

FUNDING

This work was supported by funding from Taiho Pharmaceutical under a study contract. The Comprehensive Support Project for Oncology Research of the Public Health Research Foundation in Japan conducted this study. Taiho Pharmaceutical had no role in the design of the study. The company did not participate in the collection, management, analysis or interpretation of the data; the preparation, review or approval of the manuscript; or the decision to submit the manuscript for publication.

Conflict of interest: none declared.

DATA AVAILABILITY

The data underlying this article will be shared on reasonable request to the corresponding author.

Author contributions

Hiroyuki Sakurai: Conceptualization; Formal analysis; Investigation; Methodology; Resources; Writing—original draft. Yasushi Goto: Conceptualization; Investigation; Methodology; Resources; Writing—review and editing. Kiyotaka Yoh: Conceptualization; Investigation; Methodology; Resources; Writing—review and editing. Kazuya Takamochi: Conceptualization; Investigation; Methodology; Resources; Writing—review and editing. Takehiro Shukuya: Conceptualization; Investigation; Methodology; Resources; Writing—review and editing. Tomoyuki Hishida: Conceptualization; Investigation; Methodology; Resources; Writing—review and editing. Masahiro Tsuboi: Conceptualization; Investigation; Methodology; Resources; Writing—review and editing. Koichi Yoshida: Resources; Writing—review and editing. Yasuhisa Ohde: Resources; Writing—review and editing. Sakae Okumura: Resources; Writing—review and editing. Masataka Taguri: Data curation; Formal analysis; Resources; Writing—review and editing. Hideo Kunitoh: Conceptualization; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Writing—original draft.

Reviewer information

European Journal of Cardio-Thoracic Surgery thanks Mohammad Behgam Shadmehr and Michael John Shackcloth for their contribution to the peer review process of this article.

REFERENCES

ABBREVIATIONS

 
• EGFR

  Epidermal growth factor receptor

 
• GGA

  Ground-glass area

 
• HR

  Hazard ratio

 
• HRCT

  High-resolution computed tomography

 
• JCOG

  Japan Clinical Oncology Group

 
• NSCLC

  Non-small-cell lung cancer

 
• OS

  Overall survival

 
• TNM

  Tumour, node, metastasis

 
• WHO

  World Health Organization