Prevalence of lung tumors in patients with esophageal squamous cell carcinoma and vice versa: a systematic review and meta-analysis

Purpose Recent reports suggest an increased prevalence of lung second primary tumors (LSPTs) in esophageal squamous cell carcinoma (ESCC) patients and vice versa. However, the exact prevalence of SPTs remains unclear and screening for these SPTs is currently not routinely performed in western countries. We aimed to report on the prevalence of LSPTs in patients with ESCC and esophageal second primary tumors (ESPTs) in patients with lung cancer (LC). Methods Databases were searched until 25 March 2021 for studies reporting the prevalence of LSPTs in ESCC or vice versa. Pooled prevalences with 95% confidence intervals (CI) of SPTs were calculated with inverse variance, random-effects models and Clopper–Pearson. Results Nineteen studies in ESCC patients and 20 studies in LC patients were included. The pooled prevalence of LSPTs in patients with ESCC was 1.8% (95% CI 1.4–2.3%). For ESPTs in LC patients, the pooled prevalence was 0.2% (95% CI 0.1–0.4%). The prevalence of LSPTs in ESCC patients was significantly higher in patients treated curatively compared to studies also including palliative patients (median 2.5% versus 1.3%). This difference was consistent for the ESPT prevalence in LC patients (treated curatively median 1.3% versus 0.1% for all treatments). Over 50% of the detected SPTs were squamous cell carcinomas and were diagnosed metachronously. Conclusion Patients with ESCC and LC have an increased risk of developing SPTs in the lungs and esophagus. However, the relatively low SPT prevalence rates do not justify screening in these patients. Further research should focus on risk stratification to identify subgroups of patients at highest risk of SPT development. Supplementary Information The online version contains supplementary material available at 10.1007/s00432-022-04103-0.


Introduction
Over half a million esophageal cancers and 2 million lung cancers (LC) were diagnosed worldwide in 2018 (Arnold et al. 2020;Bray et al. 2018;Lu et al. 2019). The major risk factor for esophageal squamous cell carcinoma (ESCC) and LC is tobacco smoking (Freedman et al. 2016). The prognosis of both cancers remains poor, although the 5-year survival rate has improved to approximately 22% for ESCC in 2018 and 23% for LC in 2020 (Putten et al. 2018;State of Lung Cancer 2021). The poor survival rates of patients with ESCC and LC could partially be explained by the occurrence of second primary tumors (SPTs) (Lu et al. 2019;Ven et al. 2019Ven et al. , 2020. For patients with ESCC, the occurrence of SPTs is frequently explained by the theory of field cancerization (Slaughter et al. 1953). This theory states that chronic exposure of the epithelium surrounding the primary tumor to carcinogens, especially tobacco, can lead to (pre)malignant changes of the epithelium. Most SPTs in patients with ESCC are located in the upper aero-digestive tract, especially in the head and neck region and lungs (Ven et al. 2020). Large incidence differences for both ESCC and LC exist worldwide, with high incidence rates of both cancers reported in Eastern Asia (Bray et al. 2018). However, little is known regarding the prevalence of LSPTs and ESPTs in this patient population, especially in non-Asian countries. Moreover, the potential yield and benefit of screening for SPTs in patients with ESCC and LC remains unclear.
Nowadays, screening for LSPTs in patients with ESCC and esophageal second primary tumors (ESPTs) in patients with LC is not routinely implemented in Western countries (guideline non-small cell lung cancer 2021; Guideline esophageal cancer 2021; Guideline small cell lung carcinoma 2021). According to current Asian guidelines, a trachea-bronchoscopy to detect SPTs is advised during the diagnostic workup in all patients with ESCC with chronic alcohol and tobacco consumption (Lordick et al. 2016;Muro et al. 2019). The Dutch guidelines suggest screening for LSPTs in ESCC patients may be considered and does not mention screening for ESPTs in patients with LC (Guideline esophageal cancer 2021).
The primary objective of this systematic review and meta-analysis is to investigate the prevalence of LSPTs in patients with ESCC and the prevalence of ESPTs in patients with LC. The secondary objectives are to assess the tumor stage of SPTs and time interval between the primary cancer diagnosis and detection of SPTs.

Search strategy
The databases PubMed, Embase, Medline, Cochrane Central, Google Scholar, and Web of Science were searched by two independent investigators (L.T. and S.V.) until 25 March 2021. The systematic search contained keywords for second/ multiple primary tumor, esophageal cancer and lung cancer. No time restrictions were set. The search was performed in collaboration with the medical library of the Erasmus University Rotterdam, the Netherlands. The complete search strategy is available in Supplementary Appendix 1. In addition, reference lists of included studies were searched to identify additional relevant studies.

Study inclusion
Studies that reported the proportion of LSPTs (of all histological types) in patients with ESCC or the proportion of ESPTs (both ESCC and esophageal adenocarcinoma) in patients with LC were included. Studies without original data, case reports, non-human and non-English studies were excluded. Two independent investigators (L.T. and S.V.) screened titles and abstracts followed by full texts of potentially eligible articles identified by the search strategy. In case of any disagreement, a consensus was reached through discussion (with L.T., S.V., and A.K.). The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart was used to create an overview of the data screening process (Moher et al. 2009).

Data extraction and quality assessment
The extracted information from each study included: study characteristics (author, year of publication, study country, design, and setting) and patient characteristics (gender, number of patients with ESCC and LSPTs, number of patients with LC and ESPTs, time interval between the primary cancer diagnosis and detection of SPTs, tumor stage, histopathology, and treatment). The methodological quality of each study was assessed with the Newcastle-Ottawa scale for quality assessment for cohort studies (Wells et al. 2000). Funnel plots and Egger tests were used to assess the risk of publication bias (Duval and Tweedie 2000).

Outcomes and definitions
The primary outcomes were (1) the pooled prevalence of LSPTs in patients with ESCC and (2) the pooled prevalence of ESPTs in patients with LC. Secondary outcomes included the tumor stage of SPTs and the time from the diagnosis of the primary cancer to the detection of an SPT. The criteria for SPTs from Warren and Gates were used; an SPT must be (1) a malignant tumor based on histopathological assessment, (2) separated from the primary cancer by normal mucosa, and (3) the possibility of the SPT being a recurrence or metastasis from the primary cancer must be ruled out (Warren 1932). The time to the detection of SPTs was classified as a tumor in the history before the diagnosis of ESCC or LC and synchronous and metachronous SPTs (Cahan et al. 1976). Synchronous SPTs were defined as the detection of an SPT within 6 months of the diagnosis of the primary tumor (this may be referred to as simultaneous). Metachronous SPTs were defined as the detection of an SPT at least 6 months after the diagnosis of the primary tumor.

Data analysis
For the meta-analysis, the SPT prevalence was calculated for each study as the number of SPTs divided by the number of the patient population in that specific study. The heterogeneity between included studies was assessed using the inconsistency index (I 2 ). The incidence of both ESCC and LC differs strongly worldwide, with the highest incidence rates of both cancers reported in Eastern Asia (Bray et al. 2018). Therefore, the random-effects model with inverse variance was used to calculate the pooled prevalence and 95% confidence intervals (CI) were calculated with Clopper-Pearson. Excessive influence of individual studies on the pooled prevalence was investigated in sensitivity analyses. Standardized incidence ratios (SIRs) of the included studies were extracted for a comparison with the risk in the general population to develop lung cancer or esophageal cancer. Data were presented as counts with percentages. Analyses were performed in R version 4.1.1 (The R Foundation Statistical Computing, Vienna, Austria) with meta version 4.18-2 and metafor version 3.0-2. All tests were performed two-sided and P < 0.05 was considered significant.

Study selection and quality assessment
The literature search identified 13,594 records (shown in Fig. 1). After removing duplicates, 7,782 articles were assessed for titles and abstracts, of which 171 articles were potentially eligible. After full-text reviewing, 39 studies were included in this systematic review and meta-analysis. The quality assessment according to the Newcastle-Ottawa Scale of included studies is shown in Supplementary Table 1.
The funnel plots and Egger tests showed no proof of publication bias for the prevalence of LSPTs in patients with ESCC (P = 0.11) and the prevalence of ESPTs in patients with LC (P = 0.16) ( Supplementary Fig. 1).

Characteristics of LC
The tumor stage of LC was reported in five studies (Abdel-Rahman and Cheung 2017; Faehling et al. 2018;Reinmuth et al. 2013;Takigawa et al. 2006;Fink-Neuboeck et al. 2020); however, none of these studies reported the number of ESPTs for each LC tumor stage. In six studies, only patients with LC treated with curative intent were included. Haraguchi et al. (2007), Komatsu et al. (2019), Son et al. (2013), Takigawa et al. (2006), Shimizu et al. (2001), andFink-Neuboeck et al. (2020) reported treatments for LC were surgery (n = 61,356) and chemo-or-radiotherapy (n = 108,961). Table 3 shows the studies that reported SIRs for the risk of SPTs, compared to the risk of esophageal or LC in the general population (Ven et al. 2020;Hu et al. 2015;Chen et al. 2019;Chuang et al. 2008Chuang et al. , 2010Abdel-Rahman and Cheung 2017;Levi et al. 1999;Su et al. 2017;Teppo et al. 2001). In all four studies in ESCC patients, a significantly increased risk for LSPTs was reported compared to the general population (Ven et al. 2020;Hu et al. 2015;Chen et al. 2019;Chuang et al. 2008). In five studies performed in patients with LC, SIRs ranging from 1.45 to 2.40 were reported. The study of Abdel-Rahman and Cheung 2017 reported a significantly increased risk for ESPTs in patients with LC, whereas the smaller studies of Su et al. 2017 and did not Levi et al. 1999.

Discussion
To the best of our knowledge, this is the first systematic review reporting on the prevalence of SPTs in the esophagus and lungs in patients with ESCC and LC. We found a pooled prevalence of LSPTs of 1.8% in patients with ESCC and a prevalence of ESPTs of 0.2% in patients with LC. More than 50% of the detected SPTs were squamous cell carcinomas and were diagnosed metachronously. The prevalence rates of SPTs in patients with ESCC and LC in this meta-analysis are most likely an underestimation of the actual prevalence of LSPTs in patients with ESCC and vice versa for the following reasons. First, the overall survival rates of patients with ESCC and LC remain poor, although they have increased during the recent decades (Lu et al. 2019;Putten et al. 2018). In 23 of 39 studies, patients treated with palliative intent were also included, while these patients are known to have a median survival of 22 weeks for ESCC and 20 weeks for LC (Lu et al. 2019;Putten et al. 2018). This short life span after the diagnosis of the primary tumor limits the risk for SPT development, while patients treated with curative intent are known to have better survival rates and, therefore, the cumulative risk of SPT development increases over time. This survival bias is also supported by our finding that patients treated with curative intent are significantly more at risk of developing LSPTs and ESPTs than patients who received palliative care. One can hypothesize that the cumulative SPT risks increase in the future, if treatment and survival rates of patients with ESCC and LC may continue to rise.
Second, we found a higher prevalence of LSPTs in patients with ESCC than the prevalence of ESPTs in patients with LC. This difference could be partly explained by the differential use of the positron emission Third, almost all included studies were performed retrospectively, which hampers accurate differentiation between LSPTs and lung metastases of primary ESCC. This difficulty resulted in conservative definitions of LSPTs, e.g., one study choose to exclude all lung squamous cell carcinoma detected within the first 5 years after the diagnosis of ESCC as potential SPTs (Hu et al. 2015) and another only included squamous cell lung carcinoma as LSPTs when the tumors showed clear histologic differences (Motoyama et al. 2003).
In our systematic review, nine included studies reported standardized incidence ratios (SIRs) to develop LSPTs or LSPTs. Most of these studies reported increased SIRs, supporting that SPT prevalence rates found in this study exceed the risk to develop EC and LC in the general population. However, for an adequate comparison with the risk among the general population, matching of all individual patient data of the included studies for parameters, including age, gender, comorbidities, follow-up time and alcohol and tobacco use would be essential.
The SPT prevalence rates found in this meta-analysis currently do not support screening for LSPTs and ESPTs. Future research should focus on identification of subgroups of patients with ESCC and LC with the highest risks for SPT development. Although evidence is limited, patient characteristics with the highest risk for SPTs that can be considered are for example males with chronic tobacco use and early and curable primary tumors. In these patients, the occurrence of SPTs can have major consequences for treatment and prognosis, and screening might potentially be beneficial. Moreover, geographic differences in the incidence of ESCC, LC, and SPTs are an important differentiator in the process of identification of patients with highest risks to develop SPTs. Another issue with regard to screening that needs to be addressed is the optimal timing to screen for SPTs in these patients. This needs to be balanced, between as early as possible to detect SPT at an early and curable stage on one hand and screening of selected patients with improved survival rates on the other hand.
Recently, a large-scale screening study was performed to detect lung cancers among a population of heavy (ex) smokers (Koning et al. 2020). In this study, patients underwent a minimum of 10 years of screening and follow-up with CTs at baseline, year 1, year 3, and year 5.5. The incidence of LC was 5.6%, and screening successfully reduced LC-related mortality. With our findings, combined with the fact that 80-90% of ESCC patients are heavy (ex) smokers (Gruner et al. 2020), one might hypothesize that a subgroup of patients with ESCC would also potentially benefit from CT screening during the ESCC follow-up to detect LSPTs.
Although this systematic review included all available studies reporting on the prevalence of LSPTs and ESPTs, several limitations need to be discussed: (1) different definitions for the diagnosis and timing for SPTs were used. Synchronous and metachronous SPTs were lumped together as subsequent SPTs in nine studies (Hu et al. 2015;Nagasawa et al. 2000;Chen et al. 2019;Coyte et al. 2014;Komatsu et al. 2019;Levi et al. 1999;Takigawa et al. 2006;Shimizu et al. 2001;Fink-Neuboeck et al. 2020) and varying definitions were used for synchronous and metachronous in eight studies (Kumagai et al. 2001;Abdel-Rahman and Cheung 2017;Duchateau and Stokkel 2005;Haraguchi et al. 2007;Kawahara et al. 1998;Reinmuth et al. 2013;Su et al. 2017;Teppo et al. 2001); (2) the retrospective study design with limited information regarding the detection method of SPTs and lack of long-term follow-up data in most included studies; (3) both ESCC and LC often remain asymptomatic for a long time and, therefore, are frequently detected in advanced stages; (4) high heterogeneity between the included studies. These limitations in the methodology of included studies resulted in rather low prevalence rates of SPTs.
In conclusion, this meta-analysis showed that patients with ESCC and LC have an increased risk of developing SPTs in the lungs and esophagus. However, based on the rather low SPT prevalence rates found in this systematic review, screening cannot be recommended. Further research focusing on risk stratification for subgroups of patients with ESCC and LC might reveal subgroups with higher risks, potentially making screening more worthwhile.