Tumor Biol. DOI 10.1007/s13277-014-2330-1
RESEARCH ARTICLE
MicroRNAs as ideal biomarkers for the diagnosis of lung cancer Zhiqiang Guo & Chuncheng Zhao & Zheng Wang
Received: 1 June 2014 / Accepted: 7 July 2014 # International Society of Oncology and BioMarkers (ISOBM) 2014
Abstract Lung cancer (LC) is one of the most prevalent causes of cancer death with a high mortality rate worldwide. While various sets of microRNAs (miRNAs) have been found to be highly sensitive and specific biomarkers for the early diagnosis of LC (the first word of abstract), conflicting results on their diagnostic accuracy are still present in individual studies. Thus, we aimed to conduct a systematic review and meta-analysis of the published literature to comprehensively assess the diagnostic value of miRNAs for predicting LC. The sensitivity and specificity of each included study were used to plot the summary receiver operator characteristic (SROC) curve and to calculate the area under the SROC curve (AUC). All analyses were performed using the Stata 12.0 software. Twenty-six articles were involved in our metaanalysis, 18 of which focused on single miRNA assays and 15 on multiple miRNA assays. For single miRNA profiling, the pooled parameters calculated from all studies are as follows: sensitivity (SEN), 0.72; specificity (SPE), 0.74; positive likelihood ratio (PLR), 2.7; negative likelihood ratio (NLR), 0.39; and diagnostic odds ratio (DOR), 7. For multiple miRNA profiling, the pooled estimates for the overall studies are as follows: SEN, 0.81; SPE, 0.84; PLR, 4.9; NLR, 0.23; and DOR, 22, which are significantly better than the diagnostic performance of the single miRNA profiling. In addition, subgroup analyses based on sample types suggested that blood-based multiple miRNA assays were more accurate than non-blood-based studies. In conclusion, the current metaanalysis shows that multiple miRNA assays were more accurate in diagnosing LC than single miRNA assays. However, further large-scale investigations are urgently Zhiqiang Guo and Chuncheng Zhao contributed equally to this paper. Z. Guo : C. Zhao : Z. Wang (*) The Department of Thoracic Surgery, Shanghai Putuo District Central Hospital, Shanghai 200062, China e-mail: [email protected]
needed to confirm our results and verify the feasibility of routine clinical utilization. Keywords MicroRNAs . Lung cancer . Diagnostic value . Meta-analysis
Introduction Lung cancer is one of the most prevalent causes of cancer death and has a high mortality rate worldwide [1, 2]. Nonsmall cell lung cancer (NSCLC) accounts for about 80 % of lung cancer (LC) cases, and small cell lung carcinoma (SCLC) almost comprises the other 20 % [3]. Due to sluggish progress over the past decades in the development of early diagnosing methods, there has been no significant decrease in the high mortality rate of LC, which is largely due to late-stage diagnoses [4]. Most LC cases, particularly at early stages (I and II), are usually asymptomatic until they reach advanced stages and thus are difficult to be diagnosed early [5]. More than 75 % of LC cases are diagnosed only after having locally advanced or metastasized because we lack sensitive and specific biomarkers for early diagnosis [2]. The 5-year survival rate of LC patients at early stages (I and II) can be as high as 83 % after receiving effective treatments, but for LC patients at advanced stages (III and IV), this rate dramatically decreases to 14 % [6]. Since substantial survival advantage is conferred by an early diagnosis, extensive efforts have been made to identify novel markers, which would reduce mortality and improve therapeutic outcomes. Currently, the diagnostic gold standard for LC involves identifying morphological characteristics of malignant cells through invasive methods, such as bronchoscopy, chest X-ray, computed tomography (CT), and positive emission tomography (PET) [7–10]. Bronchoscopy, which excels at detecting centrally occurring LC, is not routinely performed despite its
Tumor Biol.
invasive and costly nature [7]. Although chest X-rays and CT screenings appear promising given that they can detect LC at early stages, they have relatively low specificities and high false positive rates [9, 10]. The high rate of false positives may cause anxiety and even lead to unnecessary biopsies or surgeries that carry risks for patients with benign diseases. In addition, its potential benefits may be overwhelmed by the hazards of the associated radiation. Finally, while sensitivity and specificity of PET are both high, it is expensive and does not avoid surgical resection [8]. These facts have compelled researchers to seek reliable, noninvasive, and cost-effective confirmatory assays and novel biomarkers to noninvasively diagnose lung tumors, which would reduce the overdiagnosis and facilitate the implementation of conventional reference gold standard tests. Over the last decade, the search for novel biomarkers for diagnosis is becoming a hot research field in LC research. The desired biomarker must be easily accessible through noninvasive methods and sensitive enough to differentiate LC patients at early stages from cancerfree individuals [11]. Several currently available potential biomarkers, such as cytokeratin-19 fragment (CYFRA21-1) [12], carcinoembryonic antigen (CEA) [13], neuron-specific enolase (NSE) [14], cancer antigen 125 (CA-125), and cancer antigen 19 (CA-19) [15], provide the potential to comprehensively detect LC without the need to perform a surgical procedure or biopsy. However, few of these markers have sufficient diagnostic power to be used exclusively. Due to their limited sensitivity and specificity, there are considerable barriers to their wide use in clinical contexts. The discovery of microRNAs (miRNAs), which are a group of noncoding small RNAs, has opened up new perspectives for tumor diagnosis and provides a novel approach for the initial screening of tumors, including LC [16]. It has been widely accepted that miRNAs may act as tumor suppressors and that aberrant miRNA expression has been proved to be associated with tumorigenesis and the regulation of critical cellular processes, including apoptosis and cell growth [17]. Accumulating evidence has reported that dysregulated miRNA expression may contribute to tumor initiation and progression in human cancers [18–21]. Moreover, the high stability of miRNAs constitutes an enormous advantage from a clinical diagnostic point of view [22]. It allows an efficient isolation from clinical specimens including tissue, sputum, plasma, and serum. The above advantages of miRNA expression detection have made them ideally suited to serve as noninvasive biomarkers for the early diagnosis of LC. Several recent studies have indicated that various sets of miRNAs can be used as highly sensitive and specific markers for the early detection of LC. However, conflicting results are
still present in individual studies, which often identified different miRNA signatures. Therefore, a systematic analysis of these data may be valuable to further explore the clinical applicability of miRNAs as biomarkers for the diagnosis of LC.
Materials and methods Literature search This meta-analysis was conducted according to guidelines for diagnostic meta-analysis [23]. PubMed, Embase, Chinese National Knowledge Infrastructure (CNKI), and Chinese Biomedical Literature Database (CBM) were searched up to April 1, 2014, using the search terms (“lung neoplasm” OR “lung cancer” OR “lung malignancy”) AND (“microRNAs” OR “miRNA”) AND (“diagnosis” OR “sensitivity” OR “specificity” OR “ROC curve”) without language restrictions. In addition, the reference lists of relevant reviews were manually searched to obtain additional articles. Inclusion and exclusion criteria To be eligible for inclusion in this meta-analysis, studies had to fulfill the following criteria: (1) concerned the diagnostic potential of miRNAs for LC, (2) used the diagnostic gold standard for confirming LC, and (3) provided sufficient data for the construction of two-by-two tables, including true positive (TP), false positive (FP), true negative (TN), and false negative (FN). Exclusion criteria were (1) not related to the diagnostic values of miRNAs for LC, (2) reported duplicate data from other studies, and (3) took the form of letters, editorials, case reports, or reviews. Data extraction Two investigators independently extracted the following data from all eligible studies using a standardized form: (1) basic characteristics of studies, including name of the first author, year of publication, country of origin, mean age, male ratio, patient spectrum, source of control, methods of miRNA detection, and sample type and (2) diagnostic outcomes, including the number of patients with sensitivity, specificity, TP, FP, FN, and TN. Quality assessment Two investigators independently assessed the qualities of the included studies using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) criteria [24]. The QUADAS-2 tool is comprised of four key domains: patient selection, index test, reference standard, and flow and timing,
Tumor Biol.
and it uses seven questions to evaluate the quality of diagnostic accuracy studies. Each question is answered with “yes,” “no,” or “unclear.” An answer of yes means that the risk of bias can be judged low, whereas an answer of no or unclear means that there is a potential risk of bias.
of the included studies. Since publication bias is a concern for meta-analyses of diagnostic accuracy studies, Deeks’ funnel plot asymmetry test was performed, with P
RESEARCH ARTICLE
MicroRNAs as ideal biomarkers for the diagnosis of lung cancer Zhiqiang Guo & Chuncheng Zhao & Zheng Wang
Received: 1 June 2014 / Accepted: 7 July 2014 # International Society of Oncology and BioMarkers (ISOBM) 2014
Abstract Lung cancer (LC) is one of the most prevalent causes of cancer death with a high mortality rate worldwide. While various sets of microRNAs (miRNAs) have been found to be highly sensitive and specific biomarkers for the early diagnosis of LC (the first word of abstract), conflicting results on their diagnostic accuracy are still present in individual studies. Thus, we aimed to conduct a systematic review and meta-analysis of the published literature to comprehensively assess the diagnostic value of miRNAs for predicting LC. The sensitivity and specificity of each included study were used to plot the summary receiver operator characteristic (SROC) curve and to calculate the area under the SROC curve (AUC). All analyses were performed using the Stata 12.0 software. Twenty-six articles were involved in our metaanalysis, 18 of which focused on single miRNA assays and 15 on multiple miRNA assays. For single miRNA profiling, the pooled parameters calculated from all studies are as follows: sensitivity (SEN), 0.72; specificity (SPE), 0.74; positive likelihood ratio (PLR), 2.7; negative likelihood ratio (NLR), 0.39; and diagnostic odds ratio (DOR), 7. For multiple miRNA profiling, the pooled estimates for the overall studies are as follows: SEN, 0.81; SPE, 0.84; PLR, 4.9; NLR, 0.23; and DOR, 22, which are significantly better than the diagnostic performance of the single miRNA profiling. In addition, subgroup analyses based on sample types suggested that blood-based multiple miRNA assays were more accurate than non-blood-based studies. In conclusion, the current metaanalysis shows that multiple miRNA assays were more accurate in diagnosing LC than single miRNA assays. However, further large-scale investigations are urgently Zhiqiang Guo and Chuncheng Zhao contributed equally to this paper. Z. Guo : C. Zhao : Z. Wang (*) The Department of Thoracic Surgery, Shanghai Putuo District Central Hospital, Shanghai 200062, China e-mail: [email protected]
needed to confirm our results and verify the feasibility of routine clinical utilization. Keywords MicroRNAs . Lung cancer . Diagnostic value . Meta-analysis
Introduction Lung cancer is one of the most prevalent causes of cancer death and has a high mortality rate worldwide [1, 2]. Nonsmall cell lung cancer (NSCLC) accounts for about 80 % of lung cancer (LC) cases, and small cell lung carcinoma (SCLC) almost comprises the other 20 % [3]. Due to sluggish progress over the past decades in the development of early diagnosing methods, there has been no significant decrease in the high mortality rate of LC, which is largely due to late-stage diagnoses [4]. Most LC cases, particularly at early stages (I and II), are usually asymptomatic until they reach advanced stages and thus are difficult to be diagnosed early [5]. More than 75 % of LC cases are diagnosed only after having locally advanced or metastasized because we lack sensitive and specific biomarkers for early diagnosis [2]. The 5-year survival rate of LC patients at early stages (I and II) can be as high as 83 % after receiving effective treatments, but for LC patients at advanced stages (III and IV), this rate dramatically decreases to 14 % [6]. Since substantial survival advantage is conferred by an early diagnosis, extensive efforts have been made to identify novel markers, which would reduce mortality and improve therapeutic outcomes. Currently, the diagnostic gold standard for LC involves identifying morphological characteristics of malignant cells through invasive methods, such as bronchoscopy, chest X-ray, computed tomography (CT), and positive emission tomography (PET) [7–10]. Bronchoscopy, which excels at detecting centrally occurring LC, is not routinely performed despite its
Tumor Biol.
invasive and costly nature [7]. Although chest X-rays and CT screenings appear promising given that they can detect LC at early stages, they have relatively low specificities and high false positive rates [9, 10]. The high rate of false positives may cause anxiety and even lead to unnecessary biopsies or surgeries that carry risks for patients with benign diseases. In addition, its potential benefits may be overwhelmed by the hazards of the associated radiation. Finally, while sensitivity and specificity of PET are both high, it is expensive and does not avoid surgical resection [8]. These facts have compelled researchers to seek reliable, noninvasive, and cost-effective confirmatory assays and novel biomarkers to noninvasively diagnose lung tumors, which would reduce the overdiagnosis and facilitate the implementation of conventional reference gold standard tests. Over the last decade, the search for novel biomarkers for diagnosis is becoming a hot research field in LC research. The desired biomarker must be easily accessible through noninvasive methods and sensitive enough to differentiate LC patients at early stages from cancerfree individuals [11]. Several currently available potential biomarkers, such as cytokeratin-19 fragment (CYFRA21-1) [12], carcinoembryonic antigen (CEA) [13], neuron-specific enolase (NSE) [14], cancer antigen 125 (CA-125), and cancer antigen 19 (CA-19) [15], provide the potential to comprehensively detect LC without the need to perform a surgical procedure or biopsy. However, few of these markers have sufficient diagnostic power to be used exclusively. Due to their limited sensitivity and specificity, there are considerable barriers to their wide use in clinical contexts. The discovery of microRNAs (miRNAs), which are a group of noncoding small RNAs, has opened up new perspectives for tumor diagnosis and provides a novel approach for the initial screening of tumors, including LC [16]. It has been widely accepted that miRNAs may act as tumor suppressors and that aberrant miRNA expression has been proved to be associated with tumorigenesis and the regulation of critical cellular processes, including apoptosis and cell growth [17]. Accumulating evidence has reported that dysregulated miRNA expression may contribute to tumor initiation and progression in human cancers [18–21]. Moreover, the high stability of miRNAs constitutes an enormous advantage from a clinical diagnostic point of view [22]. It allows an efficient isolation from clinical specimens including tissue, sputum, plasma, and serum. The above advantages of miRNA expression detection have made them ideally suited to serve as noninvasive biomarkers for the early diagnosis of LC. Several recent studies have indicated that various sets of miRNAs can be used as highly sensitive and specific markers for the early detection of LC. However, conflicting results are
still present in individual studies, which often identified different miRNA signatures. Therefore, a systematic analysis of these data may be valuable to further explore the clinical applicability of miRNAs as biomarkers for the diagnosis of LC.
Materials and methods Literature search This meta-analysis was conducted according to guidelines for diagnostic meta-analysis [23]. PubMed, Embase, Chinese National Knowledge Infrastructure (CNKI), and Chinese Biomedical Literature Database (CBM) were searched up to April 1, 2014, using the search terms (“lung neoplasm” OR “lung cancer” OR “lung malignancy”) AND (“microRNAs” OR “miRNA”) AND (“diagnosis” OR “sensitivity” OR “specificity” OR “ROC curve”) without language restrictions. In addition, the reference lists of relevant reviews were manually searched to obtain additional articles. Inclusion and exclusion criteria To be eligible for inclusion in this meta-analysis, studies had to fulfill the following criteria: (1) concerned the diagnostic potential of miRNAs for LC, (2) used the diagnostic gold standard for confirming LC, and (3) provided sufficient data for the construction of two-by-two tables, including true positive (TP), false positive (FP), true negative (TN), and false negative (FN). Exclusion criteria were (1) not related to the diagnostic values of miRNAs for LC, (2) reported duplicate data from other studies, and (3) took the form of letters, editorials, case reports, or reviews. Data extraction Two investigators independently extracted the following data from all eligible studies using a standardized form: (1) basic characteristics of studies, including name of the first author, year of publication, country of origin, mean age, male ratio, patient spectrum, source of control, methods of miRNA detection, and sample type and (2) diagnostic outcomes, including the number of patients with sensitivity, specificity, TP, FP, FN, and TN. Quality assessment Two investigators independently assessed the qualities of the included studies using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) criteria [24]. The QUADAS-2 tool is comprised of four key domains: patient selection, index test, reference standard, and flow and timing,
Tumor Biol.
and it uses seven questions to evaluate the quality of diagnostic accuracy studies. Each question is answered with “yes,” “no,” or “unclear.” An answer of yes means that the risk of bias can be judged low, whereas an answer of no or unclear means that there is a potential risk of bias.
of the included studies. Since publication bias is a concern for meta-analyses of diagnostic accuracy studies, Deeks’ funnel plot asymmetry test was performed, with P
