Lung cancer in idiopathic pulmonary fibrosis: A systematic review and meta-analysis

Background There are many epidemiological pieces of evidence that show IPF patients have the highest risk of lung cancer. We conducted a systematic review of all published data to define the characteristics of lung cancer that develops in IPF by performing a meta-analysis. Method This study was performed based on the PRISMA guideline. Documents gathered by searching through the Web of Sciences, Scopus, PubMed/Medline, OVID, and COCHRANE databases which published before 03/25/2018 that related to lung cancer in IPFs’ patients. Articles were searched using standard keywords as well as Mesh and Mesh Entry and all probabilistic combinations of words using Boolean operators. Data searching, extracting and quality appraising were done by two researchers, independently. At last, Random-effects size based on Cochrane test and I2 were used. The review protocol has been registered in PROSPERO with ID: CRD42018094037. Results Based on the meta-analysis conducted in 35 (0.18%) included studies, the total sample size of patients with IPF was estimated 131947 among whom 6384 had LC. The total rate of LC prevalence in IPF patients was estimated to be 13.54% (95% CI: 10.43–17.4) that was significantly 9 times higher in men vs. Women and smoker vs. non-smoker. Highest to lowest prevalence of cellular (histological) subtypes of lung cancer in IPF were SQCC (37.82%), ADC (30.79%), SmCC (20.48%), LCC (5.21%), and ADQC (4.81%), respectively. The highest and lowest stage of lung cancer in IPF patients was estimated at III and II, respectively. The highest involvement location of lung cancer in IPF patients was in the Peripheral. Also, the prevalence of the tumor region involved from the highest to the lowest was estimated to be in the RLL, LLL, RUL and LUL regions. Conclusions Lung cancer in IPF, most commonly SQCC, presents in elderly heavy smokers with a male, locating in peripheral regions and the lower part of lung predominance.


Introduction
Idiopathic Pulmonary Fibrosis (IPF), known as Cryptogenic Fibrosing Alveolitis (CFA), diffuse Fibrosing Alveolitis (diff FA), and Usual Interstitial Pneumonitis (UIP) is one of the most common forms of Interstitial Lung Disease (ILD) over the years [1][2][3][4][5][6]. IPF is an unknown chronic pulmonary disease with unknown origin. It is a chronic lung disease characterized by a progressive and irreversible decline in lung function [1,7]. Due to the increasing prevalence of this disease in the United States of America, 48,000 new cases are diagnosed each year, with an annual mortality rate of 40,000 individuals (84%), which is as many as the mortality rate of breast cancer [8]. The incidence of IPF rises exponentially in the 50-85-year-old age groups (only 2-15% prevalence in patients under 50 [2,9,10]) [11][12][13][14][15].
Family history of disease (contrary to the common form, the prevalence in younger ages is higher [16][17][18]) [16,17,19,20], male gender, smoking, various types of environmental effects such as organic and inorganic dust, medical treatments, other medical disorders and microbial agents like Epstein-Barr viruses are risk factors that increase the risk of IPF [21]. Age over 50 years, dry and non-productive cough on exertion, progressive exertional dyspnea, dry inspiratory bibasilar crackles "Velcro-like" on auscultation, clubbing of the digits, hypoxemia, and abnormal results of lung function test and, constraint evidence and disturbance in gas exchange are symptoms, clinical and diagnostic features of IPF [11,14,15,22].
Among radiographic features used in screening IPF patients, chest x-ray (CXR) can be mentioned, which the diagnosis and clinical manifestations may show bilateral interstitial opacities with the possibility of occurrence in peripheral and lower lung zones. This demonstration may not be visible in 2 to 10% of patients [1,11,22]. Also, HRCT is very important and crucial due to the high sensitivity and specificity which can differentiate IPF from other ILDs [1,23,24]. According to the ATS / ERS / JRS / ALAT 2011 protocol, the HRCT is the essential test for diagnosing IPF. Reticular opacities (mostly associated with bronchiectasis), honeycombing manifested (typically with 3-10 mm, occasionally large, and usually sub-pleural with well-defined walls), ground-glass opacities (more common, but less than reticulation) and distribution, particularly is in the basal and peripheral and often scattered areas that can be the diagnosis and clinical manifestations in the HRCT test of patients with IPF [1]. Also, forced vital capacity (FVC) !50%, diffusing capacity for carbon monoxide (DLCO) ! 30% and 6-minute walk test (6MWT) distance! 150 meters in the spirometry test represent the mild-to-moderate level of IPF disease [25][26][27][28]. Likewise, in laboratory testing issued for screening patients with ILDs, including IPF, the Krebs von den Lungen 6 (KL-6) biomarker is referred, which is not widely used [29,30]. Since the diagnosis of IPF involves clinical, radiological and histopathological findings, the multi-diagnostic test increases the accuracy [ Despite previously published studies about the prevalence of LC in IPF patients in different countries (primary study) [3-6, 41, 44, 50-79], no comprehensive study and systematic review and meta-analysis have been conducted globally. One of the most important goals of systematic review and meta-analysis studies is combining the existing studies to increase sample size due to the increased number of related studies and to reduce the differences between the existing parameters and confidence interval, which ultimately leads to solving the review problems in the previous method. Certainly, such studies are a vital link between research studies and clinical decision making at the patient's bedside [80][81][82][83][84][85][86]. Considering the above mentioned cases and the prevalence, severity and extent of LC in IPF patients, as well as the presentation of the final conclusions for policy-making and correct management planning at the macro level, a systematic review of all documentation and their combination, is conducted via metaanalysis method to estimate the overall rate of LC in IPF patients and other risk factors.

Study protocol
The present study is based on the Meta-analysis of Observational Studies in Epidemiology guideline [87] and it has been conducted in 5 steps according to the PRISMA statement [88] (S1 File) including design and search strategy, a collection of articles and their systematic review, evaluation of inclusion and exclusion criteria, qualitative evaluation and statistical analysis of data. All steps were carried out by two researchers independently and, any encounters were assessed by a specialist. The review protocol has been registered in PROSPERO: International Prospective Register of Systematic Reviews (https://www.crd.york.ac.uk/ PROSPERO/) Identifier: CRD42018094037 [89,90](S2 File).

Search strategy
An advanced relevant search was conducted in international databases, such as Web of Sciences, Scopus, PubMed/Medline, OVID, and COCHRANE, to collect all of the studies which were related to LC in IPF patients. Articles were searched using standard keywords as well as Mesh and Mesh Entry and all probabilistic combinations of words using Boolean operators combined in accordance with the search syntax (S1 Table) on each database without time limit until 03/25/2018. And, a manual search was also done as reviewing the reference list of related articles. The important point in searching the databases was the high-sensitivity searching, and also the search was conducted by the researchers and a specialist which is expert in searching databases (A.R).

Selection of studies
After the end of the search, the papers were entered into the EndNote, reference management software and, after "Find References Updates", duplicates were removed. After blinding studies (hiding authors', the name of the journal and published year), each study was evaluated by two researchers independently in the screening stage, with skimming and scanning the study titles, to evaluate the inclusion and exclusion criteria and IPF and LC detection criteria (the eligibility stage). In the event of disagreement between the two researchers, the specialist researcher made the final decision (Fig 1).

Quality appraisal
After excluding irrelevant studies in the screening and eligibility stages, the quality of the final studies was examined. At this stage, the Newcastle-Ottawa Scale (NOS) [97] checklist (S3 File) was used which consists of 8 sections, and divides the studies with a scale score of 0 to 8 from poor to high quality, respectively. According to this scoring, the studies are divided into three levels of scoring: 1-Studies with a score of 5: poor quality; 2-studies with a score of 5-6: medium quality; 3-studies with a score of 7 to 8: high quality. Finally, (at the included stage), the articles that had medium to high quality were entered into the next stage (Fig 1).

Data extraction
At first, a checklist was designed according to the aims of the study and studying other available resources. and Location (Peripheral and Central), which were extracted by two independent researchers and blind for the name of the author, institute and journal. In necessary cases, further information and raw data were requested by contacting the author (the first author or responsible or the authors' department).

Statistical analysis
In each study, after considering the prevalence rate of LC in IPF patients as a binomial distribution probability, its variance was calculated by binomial distribution and for evaluating the heterogeneity of the studies Cochran test (Q) and I 2 index were used. The I 2 index less than 25% is low heterogeneity, between 75% -25% is the average heterogeneity and more than 75% are considered as heterogeneous [85,98]. According to the heterogeneity of studies (high), the random effects model has been used to combine the results of studies. Sensitivity analysis ("One Study Removed" test) was conducted to investigate the impact of each study on total results for the overall prevalence and each of the risk factors. In order to evaluate the cause of heterogeneity, the subgroup analysis was performed based on the country, and gender. The Meta-Regression model was used to determine the prevalence rate based on the year of publication. The Egger and Begg's test were evaluated to examine the publication bias (by using the Funnel Plot). Data analysis was performed using the Comprehensive Meta-Analysis Ver.2, and the significance level of the test was considered less than 0.05.

Search results and characteristics
In this systematic study, based on performed searches, 667 articles were identified and after conclusive investigation and evaluation according to the checklist, 35 (0.18%) articles[3-6, 44, 50-79](S2 Table)

Overall prevalence of LC in IPF
The overall prevalence rate of LC in patients with IPF was calculated at 13.54% (95% Confi-

Prevalence of LC in IPF based on country
Among the studies in 8 countries, the highest and lowest prevalence of LC in IPF patients was in Japan with 12 studies and Denmark with 1 study that were estimated 22.12% (95% CI: 11.35-38.64) and 5.79% (95% CI: 2.79-11.64) which were statistically significant (P = 0.000) (Fig 3).

Cumulative meta-analysis and sensitivity analysis of studies
Sensitivity analysis of LC prevalence in IPF patients and the confidence interval of each study was calculated with a 95% confidence interval, and results showed that before and after exclusion of each study there had been no significant effect on the overall prevalence rate of LC in IPF patients (Fig 4). Cumulative meta-analysis is also estimated for the overall prevalence of LC in IPF patients based on the publication year and is represented in (Fig 5).

Prevalence of smoking statue in LC-IPFs' patients
The prevalence of smoking in LC-IPF patients were estimated to be 90.

Prevalence of LC in IPF according to clinical staging
The highest and lowest Stage of LC in IPF patients were estimated in III and II with a prevalence of 30.72% (95% CI: 22.68-40.14) and 13.33% (95% CI: 8.74-19.81), respectively (P = 0.000). The highest density was observed in Stage III and IV (S13 Fig).

Publication bias
The publication bias was also evaluated by Begg and Egger's tests and was estimated at P = 0.516 and P = 0.0521, respectively. In this test, the probability of publication bias was not statistically significant (Fig 7).

Discussion
The total sample size of patients with IPF was estimated 131947 among whom 6384 had LC. The total rate of LC prevalence in IPF patients based on the meta-analysis review was estimated to be 13.54% (95% CI: 10.43-17.4) that was significantly 9 times higher in men than women. Also, the prevalence of smoking in LC-IPF patients is estimated to be 9 times higher. Highest to lowest prevalence of cellular (histological) subtypes of lung cancer in IPF were SQCC (37.82%), ADC (30.79%), SmCC (20.48%), LCC (5.21%), and ADQC (4.81%), respectively. The highest and lowest stage of lung cancer in IPF patients was estimated at III and II, respectively. The Lung cancer in IPF: A systematic review and meta-analysis highest involvement location of lung cancer in IPF patients was in the Peripheral region with highest to the lowest RLL, LLL, RUL and LUL regions respectively, that totally the highest to lowest was estimated to be in the lower, upper and middle regions.
Age and smoking status are also known to be the effective factors in the development of lung cancer in IPF patients [50,70]. Nearly every patient with prostate cancer and lung cancer (95%) has finger clubbing, while the percentage of IPF patients are approximately only 60%, which is often known as the clinical evidence of lung cancer [40,62].
Recent studies found that the progression of lung cancer in lower lobes is higher in IPF patients with [99,100]. As also found in the findings of the present study, there is a significant relationship between the involved lobes, which can be a phenomenon called "scar-cinoma" between fibrotic areas and cancer progression [101]. Although further studies are required to prove this [101][102][103][104][105].
In another study, epigenetic and genetic changes, abnormal expression of microRNAs (miRNAs, cellular and molecular aberrances), like different responses to regulatory signals, apoptosis, delay or decrease in cell-to-cell correlation along with activation of specific signal transmission pathways lead to pathogenic features of both LC-IPF. Likewise, genetic analysis has shown that harmful and deleterious mutations in A1 or A2-surfactant proteins cause familial idiopathic interstitial pneumonia and lung cancer [112][113][114].
The heterogeneity rate (I 2 ) in the present study was calculated at 90.71%, which is in the line of studies with high heterogeneity. It is assumed that the observed differences are due to various samplings and also the difference in the measured parameter in different societies [87,88,115].
According to the meta-regression, the prevalence of the LC-IPF was not statistically significant (p = 0.451) by the publication year. Even though the studies are in different countries and in years, but these findings cannot represent the reality in all countries, so further studies are needed to be conducted in this regard.

Limitations
One of the main limitations of this meta-analysis to be mentioned is the inclusion of studies with different inclusion and exclusion criteria, and there is no consensus definition of IPF expressed. Selection Bias is more discussed which can limit the generalization of these findings because the type of lung cancer in a country can be different with the other countries and could be related to descent diversities. Due to lack of resources and very few studies have investigated the survival rate and causes of mortality and the precise methods of treatment, according to the aim of the present study, we did not focus on these factors. Also, data were accessed by using Guilan University of Medical Sciences' -Iran Ministry of Health & Medical Education-VPN which some databases are not fully accessible.

Conclusion
In conclusion, the high prevalence of the LC-IPF with 13.5% is more observed in older men who smoke, and is more evident in the progression of cancer, SQCC, and SmCC, and mostly affects the peripheral regions and the lower part of the lung. Studies have been conducted in limited countries, such as Japan, Korea, and UK and USA, which the weakness of a unit study of LC-IPF in different countries investigating the factors and important risk factors and reaching to a consensus and preparing a comprehensive global database for clinical decision-making is felt and is an essential need.