Surgical outcomes of non-small-cell lung carcinoma in patients previously treated for gastric cancer

Abstract

OBJECTIVES

Although the incidence of non-small-cell lung cancer (NSCLC) as a second malignancy is increasing, the prognosis remains controversial. Therefore, the present study aimed to determine the prognosis of patients with NSCLC who had previously been treated for gastric cancer (PGC).

METHODS

The clinicopathological records of patients who underwent complete surgical resection for NSCLC in three institutions from 2000 to 2013 were retrospectively investigated.

RESULTS

A total of 4651 patients were eligible for this study: 100 (2.1%) were patients with PGC and 4551 (97.9%) were patients with NSCLC who had not previously been treated for gastric cancer (NGC). The populations of older patients (P < 0.001), males (P < 0.001), limited resection for NSCLC (P = 0.015) and non-adenocarcinoma (P = 0.024) were significantly higher in the PGC, than in the NGC group. Overall survival did not significantly differ between the PGC and NGC groups (76.4 vs 74.5% P = 0.82). Multivariate analysis revealed that more advanced age, male sex, higher serum carcinoembryonic antigen levels, more advanced clinical stage of lung cancer and nonadenocarcinoma were independent factors for a poor prognosis, whereas a history of gastric cancer was not. None of the factors associated with gastric cancer affected the survival of patients with PGC.

CONCLUSIONS

After surgical treatment for lung cancer, a history of gastric cancer treatment had low impact on survival and no factors related to gastric cancer influence the outcomes. Curative surgery for NSCLC should be recommended when previously treated gastric cancer is well controlled.

INTRODUCTION

The incidence of non-small-cell lung cancer (NSCLC) as a second malignancy is increasing [1–3] because the survival of patients with many other types of malignancy has improved. Treatment for NSCLC as a second malignancy was determined considering the general status and the type or prognosis of previous cancer. However, whether surgical resection was the appropriate treatment for these patients is unclear because previous cancer might have affected the surgical outcomes of lung cancer. Although the prognosis of NSCLC as a second primary malignancy has been studied in detail [1–8], it remains controversial. One possible reason for this is population heterogeneity in the previous studies due to a variety of previous malignancies. Little survival data are available about patients with NSCLC and prior specific malignancies have been reported. We highlighted gastric cancer because it is a common malignancy [9] and it could affect the survival rates of lung cancer because it is associated with higher fatality rates [10]. The incidence rate of a second primary tumour developing after gastric cancer is 4.2% and such second primary tumours comprise lung cancer in 28.4% of these patients [11]. A few studies [1, 11] have addressed the clinical behaviour and survival of small cohorts of patients with NSCLC who had previously been treated for gastric cancer (PGC). Thus, the clinicopathological characteristics and the prognosis of patients with PGC remain unknown.

Here, we evaluated the clinicopathological features of PGC, assessed the prognosis of PGC and determined whether a history of gastric cancer influences the prognosis of NSCLC.

MATERIALS AND METHODS

Patient population

This study included 4782 patients who underwent complete surgical NSCLC resection at Hiroshima University Hospital (Hiroshima, Japan), the National Cancer Center Hospital East (Chiba, Japan) and Juntendo University Hospital (Tokyo, Japan) between January 2000 and March 2013. We reviewed medical records and obtained clinicopathological information about gastric cancer and lung cancer for each patient. The surgical indications for primary lung cancer were discussed by each institutional cancer board. Sublobar resection was performed in cases of complete resection of the disease with appropriate surgical margins for small peripheral tumours. If lymph node metastasis was confirmed on an intraoperative frozen section of any lymph node, the procedure was converted to standard lobectomy. All stages were reclassified according the TNM classification of Malignant Tumors, 7th Edition [12]. The Institutional Review Boards at the participating hospitals approved the study and waived the requirement for the provision of written informed consent by individual patients.

Patients with NSCLC were assigned to groups according to whether they had been previously treated for gastric cancer (PGC) or not (NGC). Patients were excluded if they had incompletely resected NSCLC, missing information about treatment dates or stage of gastric cancer, stage IV gastric cancer and/or gastric cancer that was detected after surgery for lung cancer (Fig. 1). Data from the remaining 4651 patients were retrospectively analysed.

Pathological studies

Sections were fixed with 10% formalin and embedded in paraffin. Consecutive 4-µm sections were pathologically assessed using microscopy. Histological type was determined by staining with haematoxylin–eosin (H-E) and if the findings were inconclusive, the sections were immunohistochemically stained. Whether tumours were histologically second primary tumours or metastases was determined by pathologists from each institution based on immunohistochemical staining for TTF-1, CK7, CK20 SPA or Napsin A.

Statistical analysis

Data were statistically analysed using EZR (Saitama Medical Centre, Jichi Medical University; Kanda, 2012), which is a graphical user interface for R (The R Foundation for Statistical Computing, version 2.13.0). Summarized data are presented as numbers or as means ± standard deviation unless otherwise stated. Categorical and continuous variables were compared using the χ² test and an unpaired t-test, respectively. Overall survival (OS) was defined as the interval between the date of lung surgery and that of death or the last follow-up visit. OS curves were calculated using the Kaplan–Meier method and survival differences among patients were assessed using the log-rank test. OS was assessed by univariate and multivariate analyses using the Cox proportional hazards model. A P-value of <0.05 was considered statistically significant.

RESULTS

Clinical outcomes of patients with non-small-cell lung cancer who had previously been treated for gastric cancer

We assigned 100 (2.1%) and 4551 (97.9%) patients into PGC and NGC groups, respectively. The median follow-up duration was 62.3 months. A total of 952 patients died (PGC, n = 16; NGC, n = 936). The proportions of patients who were older (70.9 ± 7.6 vs 66.2 ± 9.7%, P < 0.001) and male (85 vs 61%, P < 0.001), and had limited resection of NSCLC (24 vs 15%, P = 0.015) and nonadenocarcinoma were significantly higher (39 vs 28%, P = 0.024) in the PGC, than in the NGC group. Serum CEA levels, clinical and pathological stages of lung cancer and adjuvant therapy for lung cancer did not significantly differ between the groups (Table 1).

Most patients in the PGC group had early-stage gastric cancer (78% had Stage I), and had been surgically treated for gastric cancer (73%). Only 3 patients had recurrent gastric cancer. Sixteen (16%) patients died (lung cancer, n = 1; gastric cancer, n = 9; other diseases, n = 6; Table 2). The median interval between gastric cancer and lung cancer was 3.2 (range 0–21.1) years. Thirty-two (41%) patients had undergone surgical lung cancer resection within 2 years of treatment for Stage I gastric cancer, and 31 (40%) had undergone such surgery over 5 years after treatment for gastric cancer (Fig. 2A). On the other hand, 7 (32%) and 8 (36%) patients underwent surgical resection for lung cancer within 2 and over 5 years after treatment for Stage II + III gastric cancer (Fig. 2B).

Prognosis of patients with non-small-cell lung cancer according to a history of treatment for gastric cancer

The 5-year OS rates did not significantly differ between the PGC and NGC groups (76.4 vs 74.5%, P = 0.82; Fig. 3), or between those in the two groups with Stages I and II + III gastric cancer (74.4 vs 74.5%, P = 0.83 and 83.0 vs 74.5%, P = 0.93). Univariate analysis of predictive factors for OS did not uncover a significant association with a history of gastric cancer (Table 3). Multivariate Cox analysis identified more advanced age, male sex, higher serum CEA levels and a higher clinical stage (II + III) of NSCLC as independent prognostic factors for poor OS, but not a history of gastric cancer [hazard ratio (HR) 1.17; 95% confidence interval (CI) 0.71–1.92; P = 0.528] (Table 3). We also examined predictive factors for OS in patients with PGC. Multivariate analysis did not associate OS with any of the factors related to gastric cancer, namely, pathological stage, method of treatment, adjuvant therapy and interval between gastric cancer and lung cancer (Table 4). Furthermore, OS did not significantly differ among intervals between gastric cancer and lung cancer (Table 5).

DISCUSSION

The present study found a higher ratio of older and male patients in the PGC, than in the NGC group and similar prognoses between the two groups after complete surgical resection of lung cancer (5-year OS: 76.4 vs 74.5%; P = 0.82). Furthermore, multivariate analysis showed that a history of gastric cancer had low impact on the prognosis of patients after complete NSCLC resection.

Very few reports have described the survival of patients with PGC [1, 11]. Ikeda et al. [11] have shown that the prognosis of patients with gastric cancer and a second primary cancer is more negatively influenced by the second primary tumour than by the primary gastric cancer. Our findings were consistent with these results. Although lung cancer accounted for a high ratio of second primary cancers among patients with gastric cancer, the study by Ikeda et al. included many types of malignancies such as colorectal, liver and oesophageal cancers, as well as other malignancies including lung cancers. On the other hand, Pages et al. [1] described particularly low OS rates among patients with NSCLC and a history of upper gastrointestinal malignancies. However, these studies included small sample cohorts and the prognosis of NSCLC with previous gastric cancer has remained controversial.

The present multivariate analysis found that having had gastric cancer had low impact on the survival of patients with NSCLC when gastric cancer was curatively treated. Our multivariate analysis showed that patients without previous gastric cancer had slightly poorer outcomes than patients with previous gastric cancer (HR 1.17; without versus with previous gastric cancer), for unknown reasons. Variables that were not included in the analysis might have been involved. In addition, multivariate analysis did not uncover any factors related to gastric cancer that were associated with OS in the PGC group and the findings of multivariate analysis were similar in both the PGC and the NGC groups (data not shown). Thus, if gastric cancer was considered controlled, then it appeared not to influence prognosis after surgical resection of lung cancer. Although the criteria for controllable gastric cancer were not defined herein, Stage I gastric cancer was considered controllable. The reason is that OS did not significantly differ between patients in the PGC group with Stage I gastric cancer and the NGC group. The results were similar for Stage II + III gastric cancer, but these findings should be carefully considered in light of the following. Firstly, only 22 patients had Stage II + III gastric cancer. Since survival rates were worse for patients with advanced than early-stage gastric cancer, a second cancer might occur less frequently in those with Stage II + III than Stage I gastric cancer because the follow-up term is short. Secondly, about 40% of surgical resections to treat lung cancer in the PGC group with Stage I gastric cancer were performed within 2 years after gastric cancer treatment. On the other hand, surgical resection to treat lung cancer in the PGC group with Stage II + III gastric cancer might be avoided even if lung cancer is diagnosed at the early stage after gastric cancer treatment.

Several reports [3, 13, 14] have suggested that the incidence of early-stage lung cancer is higher in patients with previous malignancy because of follow-up assessments after treatment for malignant tumours. Quadrelli et al. [3] described that patients with a previous malignancy were more frequently diagnosed at Stage I than those without a previous malignancy. One study found a 72% likelihood of developing Stage I + II lung cancer as a second primary tumour [13]. However, the present study found no significant differences in the clinical stages of lung cancer between the PGC and NGC groups. This is because about 40% patients with a history of gastric cancer had been surgically treated for lung cancer over 5 years after undergoing treatment for gastric cancer. These findings indicate that the frequency of routine follow-up was insufficient for many of these patients.

We found that the interval between gastric cancer and lung cancer did not influence the prognosis. A previous multivariate analysis [13] also found that the disease-free interval between the appearance of the first and the primary tumours does not significantly impact the survival of patients with lung cancer as a second primary malignancy. Thus, a previous controllable malignancy seems not to substantially influence the prognosis of lung cancer.

We also found a higher ratio of older and male patients in the PGC, than in the NGC group, which was consistent with published findings [1, 2, 11, 15]. Since cancer incidence increases with age, patients with a second primary malignancy are often older [1, 2, 11]. The incidence of secondary primary cancers is high in patients with lung, colorectal, hepatic and gastric cancers that are common among Japanese males [16]. Information about smoking was not available, but we are aware that smoking is an important factor in the development of lung and gastric cancers [17], and it is a potential risk factor for the development of a second tumour [18, 19].

The present study has some limitations. The retrospective design is one and another is the selection bias imposed by not including patients whose surgeons found no indication for surgically treating lung cancer due to the patient having had previous gastric cancer. Cox regression analysis did not reveal previous gastric cancer as a significant factor. Regardless, our findings suggested that if surgeons considered that a previous gastric cancer had been controllable, then it minimally influenced the prognosis of patients with lung cancer. However, the upper limit of the 95% CI was broad, and we should interpret this conclusion to allow the exclusion of only the effect of a history of gastric cancer history beyond a HR of 1.92.

In summary, we showed that a history of treatment for gastric cancer had low impact on postoperative survival after complete surgical resection of a second lung cancer and that no factor associated with gastric cancer influenced the surgical outcomes of lung cancer. Curative surgery for lung cancer should be recommended when gastric cancer has been curatively treated regardless of the interval between the onset of both types of cancer. Furthermore, patients with Stage I gastric cancer in particular might be considered suitable for inclusion in clinical trials. Patients with previous gastric cancer and other types of cancer should be prospectively investigated in the future.

Conflict of interest: none declared.

REFERENCES