Introduction

Lung cancer has a high incidence and fatality rate globally [1]. Approximately, 40% of stages I–III non-small cell lung cancer (NSCLC) patients often recur after radical resection [2,3,4,5]. The prognosis following recurrence was poor, and less than 2 years have been recorded as survival following a recurrence of primary lung cancer [4,5,6,7,8]. Studies have also shown that postoperative chemotherapy and radiotherapy can improve the PRS of patients [8, 9]. However post-recurrence survival (PRS) varied significantly among different patients. The PRS was negatively impacted by males [4, 6, 7], older age [4, 7, 10], poor motor status [5], abdominal and bone metastases [7], and poor differentiation [6, 10, 11]. Identifying risk factors affecting survival after relapse could help clinicians to make the subsequent treatment regimen. A lot of studies have explored or predicted prognostic factors for relapse-free survival in non-small cell lung cancer [12,13,14]. However, little is known about the factors affecting PRS in NSCLC patients, and further studies are needed.

Few studies have explored the association of recurrence sites with PRS in NSCLC patients, particularly adenocarcinoma patients. Some studies show that patients with local recurrence, as opposed to distant recurrence, have a better PRS [13, 15]. However, in several studies, distant recurrence did not impact PRS [5, 6, 16]. Besides, some studies show that poor PRS was seen in patients with bone metastases [7, 9, 11], liver metastases [6, 8], lung metastases [4], and abdominal organ metastases [7]. The survival time of liver metastasis was 7.8 months [7]. Median survival with brain metastases was 7–10 months [17]. The more metastatic organs, the shorter the survival time, but there was no significant difference in the survival time of patients with three or four or more metastatic organs [17]. Inaccurate identification of the initial recurrence site is the reason for the poor prognosis of patients after recurrence. However, there is limited research on the effect of each recurrence site on postoperative prognosis. Furthermore, there is no consensus on which recurrence site has a poorer PRS when NSCLC or adenocarcinoma first recurs.

Hence, this study aimed to explore the correlation between the initial recurrence site and PRS in NSCLC patients; to find out which organ metastasis of non-small cell lung cancer has the worst prognosis, especially adenocarcinoma; and to guide clinical decision-making and improve patient prognosis.

Patients and methods

Ethics statement

The study protocol (KY2019141) was reviewed and approved by the Institutional Review Board of the Yunnan Cancer Hospital. It was carried out following the fundamental principles of the Declaration of Helsinki. Informed consent was waived due to the retrospective nature of the study.

Patients

From January 2013 to December 2018, this retrospective study collected consecutively 3125 patients with pathologically confirmed stages I–III primary lung cancer in Yunnan Cancer Hospital after radical resection of lung cancer. Among them, 588 patients with recurrence were included in the analysis. A total of 2531 patients with recurrence and 6 patients who were lost to follow-up were excluded (Fig. 1).

Fig. 1
figure 1

Inclusion and exclusion flow charts

Full size image

Observation following surgery and recurrence

Chest and abdomen CT and blood tumor markers were routinely performed every 3 to 6 months within 3 years after surgery, and checkups were performed every 6 to 12 months for 4–5 years after surgery. Imaging examinations were carried out following symptoms when symptoms emerged during follow-up: positron emission tomography (PET)-CT or bone scan for bone pain, magnetic resonance imaging (MRI) of the head for headaches, contrast-enhanced CT or MRI, abdominal ultrasound, or gastrointestinal endoscopy for abdominal pain. After the identification of recurrence, any further systemic treatment was at the discretion of the multidisciplinary team.

Initial recurrence organs classification and word meanings

The initial recurrence organs were divided into the following seven subgroups: the (i) lung, (ii) brain, (iii) bone, (iv) abdomen (liver and adrenal glands), (v) pleural, (vi) lymph node, and (vii) multisite (two or more organs). PRS in patients with relapsing sites was analyzed. PRS was defined as the time to all-cause death from the first time there was evidence of relapse to the last observation period in an event-free review patient.

Statistical analysis

For continuous variables, the t-test was employed. For categorical variables, the Fisher exact test was employed. Both univariate and multivariate analyses for PRS were conducted using the Cox proportional hazard model. The variables included age, sex, BMI, surgical modality, degree of differentiation, pathological stage of AJCC, vascular invasion, bronchial invasion, pleural invasion, recurrent sites (lung, brain, bone, abdomen, pleura, lymph nodes, multisite), two recurrent sites, three or more relapsed sites, adjuvant chemotherapy, radiotherapy, and targeted therapy. For variables in the univariate analysis with P < 0.05, multivariate analysis was carried out. The Kaplan–Meier method was employed for calculating PRS. The log-rank test was performed to compare groups. P < 0.05 is regarded as significant in statistical terms. R4.2.2 (R project) was used for statistical analysis.

Results

In the end, 588 patients were included in the study following the inclusion and exclusion criteria. The median age of patients was 58.00 (33.00–85.00), and the median BMI was 22.65 (6.38–68.89) (Table 1). There were 109 cases (18.54%) of lung recurrence, 113 cases (19.22%) of brain recurrence, 79 cases (13.44%) of bone recurrence, 28 cases (4.76%) of abdominal recurrence, 24 cases (4.08%) of pleural recurrence, 154 cases (26.19%) of multisite recurrence, 81 cases (13.78%) of lymph node recurrence, and 28 cases (4.76%) of abdominal recurrence. There were 14 cases of liver metastasis and 14 cases of adrenal metastasis among patients with recurrence of abdominal organs. The median PRS in the total population was 23.75 (range 0.03–104.97) months. The PRS for adenocarcinoma was 25.05 (0.07–104.97) months (Table 1).

Table 1 Clinicopathological characteristics of patients
Full size table

In the total population, in comparison to patients without multisite recurrence (42.2 months, 95% CI 32.24–52.10), patients with multisite recurrence (24.8 months, 95% CI 17.46–32.20) (P = 0.026) had substantially worse PRS (Figure S1G). However, patients who had lung recurrence (63.1 months, 95% CI 51.13–74.00) had better PRS than patients who did not (31.0 months, 95% 25.10–36.96) (P < 0.001) (Figure S1A). Patients with or without brain, bone, abdominal, pleural, and lymph node recurrence were not statistically significant (Figure S1).

In adenocarcinoma, in comparison to patients without pleural recurrence (46.9 months, 95% CI 35.07–58.80), patients with pleural recurrence (21.3 months, 95% CI 15.07–27.46) (P = 0.031) had substantially worse PRS (Fig. 2E). However, patients who did lung recurrence (63.1 months, 95% CI 54.36–71.78) had better RFS than those who did not (37.3 months, 95% CI 29.0–45.60) (P = 0.005) (Fig. 2A). Patients with or without brain, bone, abdomen, lymph nodes, and multisite recurrence were not statistically significant (Fig. 2). In Fig. 2H, patients with pleural recurrence (21.3 months, 95% CI 15.07–27.46) (P = 0.004) had the worst PRS.

Fig. 2
figure 2

The PRS of patients with pleural recurrence was worse than that of patients without pleural recurrence (E), and lung recurrence is the opposite (A) in adenocarcinoma. There were no significant differences in the PRS of patients with or without recurrence at other sites (B–C and F–G). Among all recurrence sites, pleural recurrence has the worst prognosis (H). PRS, post-recurrence survival

Full size image

In squamous cell carcinoma, in comparison to patients without multisite recurrence (30.0 months, 95% CI 21.84–38.16), patients with multisite recurrence (17.8 months, 95% CI 11.19–24.41; P = 0.038) had substantially worse PRS (Figure S2G). Patients with or without lung, brain, bone, abdomen, lymph nodes, and pleural recurrence were not statistically significant (Figure S2).

In multivariate analysis of the general population, lung recurrence was a protected factor for PRS (HR 0.58, 95% CI 0.40–0.82; P = 0.003), and female, BMI (≥ 24), vascular invasion, AICC > II stage, and three or more recurrent sites were poor prognostic factors for PRS (Table S1).

In multivariate analysis of the adenocarcinoma, lung recurrence was a protective factor for PRS (HR 0.62, 95% CI 0.41–0.95; P = 0.027), and pleural recurrence (HR 2.18, 95% CI 1.14–4.17; P = 0.018), vascular invasion, AICC > II stage, and three or more recurrent sites were poor prognostic factors for PRS (Table 2). In multivariate analysis of the squamous cell carcinoma, older age (≥ 60 years) and three or more recurrence sites were poor prognostic factors for PRS (Table S2).

Table 2 Univariate and multivariate analysis of recurrence survival for adenocarcinoma
Full size table

Discussion

This study found that lung recurrence was an independent prognostic factor for PRS in the general population. Still, brain recurrence, bone recurrence, abdominal recurrence, pleural recurrence, lymph node recurrence, and multiple site recurrence did not affect PRS. In adenocarcinoma, lung and pleural recurrence were independent prognostic factors for PRS. In contrast, brain recurrence, bone recurrence, abdominal recurrence, lymph node recurrence, and multiple site recurrence did not affect PRS. It was not found which single-organ recurrence was an independent prognostic factor for PRS in squamous cell carcinoma.

Previous studies have reported that recurrence at two or more sites does not affect the prognosis of lung cancer [7]. However, in this study, we divided them into two subgroups: two recurrence sites and three or more recurrence sites. Therefore, multivariate analysis showed three or more recurrence sites with statistical significance and poor prognosis. In comparison, two recurrence sites were not statistically significant in the general population, adenocarcinoma, and squamous cell carcinoma.

Different studies have found different prognoses of PRS at various recurrence sites, such as liver recurrence. Some studies have shown that RFS in patients with abdominal organ recurrence is significantly worse and frequently recurrent [7, 18]. Additionally, it has been reported that the prognosis was poor when the liver was the first abdominal organ to recur following lung cancer resection [5, 8]; according to earlier research, patients with advanced lung cancer liver metastases also had a poor prognosis due to chemotherapy [19], TKI therapy [7, 18], immunological tolerance [20], and quicker tumor growth [21]. Nevertheless, some studies have shown that liver metastases do not influence PRS [7, 9, 11].

Different investigations have found other effects of lung recurrence on PRS prognosis. One study showed that patients with lung recurrence [4, 8] had better PRS, possibly because the lung recurrence was less malignant, which was also confirmed in our study. Moreover, radiotherapy and metastasectomy at lung recurrence sites may improve PRS in patients with lung recurrence [22, 23]. However, some research shows lung recurrence has little impact on PRS [6, 9, 11].

Some studies reported that pleural recurrence has nothing to do with PRS among lung cancer patients [7]. However, they did not do a subgroup analysis. Our study found through subgroup analysis that pleural recurrence was not statistically significant in the general population and squamous cell carcinoma in our study. Our finding was consistent with previous findings [7]. However, in adenocarcinoma, pleural recurrence patients have a poorer prognosis, possibly because there are few studies on pleural metastasis in lung cancer. A larger cohort may be needed to investigate further prognostic factors affecting pulmonary adenocarcinoma pleural recurrence to improve PRS in patients with adenocarcinoma pleural recurrence.

This study belongs to a large cohort study. For the first time, the impact of initial recurrence organs on lung cancer PRS was divided into the total population, adenocarcinoma, and squamous cell carcinoma to study separately, and the initial recurrence organs were divided into seven subgroups. There were a few limitations to this study. First, because this study was retrospective and only involved one institution, selection bias could not be ruled out; second, the method of postoperative monitoring was different for each doctor. Third, this study included fewer squamous cell carcinoma patients (n = 116), and the subsequent research could consist of more squamous cell carcinoma patients to analyze which initial recurrence site had the most significant effect on PRS in their population.

In conclusion, the initial recurrence site was associated with PRS in NSCLC patients. Identification of recurrence sites could guide the subsequent treatment. In adenocarcinoma, patients with pleural recurrence have the worst prognosis, which should be followed up as soon as possible, and early intervention treatment to improve the prognosis of patients. Patients of squamous cell carcinoma should pay more attention to multisite recurrence.