Protocol for an Umbrella Review of Non-Invasive Biomarkers used to Guide Pre-diagnostic Decision Making in Lung Cancer Screening (The BIOLUX-Umbrella Study)

Background

Lung cancer represents the foremost cause of cancer-related death, prompting significant health concerns both in Ireland and on a global scale¹. The diagnostic challenge lies in the disease's symptomatology, which often mirrors conditions like chronic obstructive pulmonary disease (COPD), asthma, and bronchitis—factors that can lead to a delayed diagnosis until advanced stages. Consequently, lung cancer is associated with a high mortality rate^1,2 underlining the critical need for early detection to improve patient outcomes.

By detecting lung cancer when it is still asymptomatic, screening tests offer a significant means to lessen the impact of this concern. The National Lung Cancer Screening Trial (NLCST) in the US currently advocates for low-dose helical computed tomography (LDCT) over chest x-rays (CXR) as the standard^1,2 for LCS. It is associated with a decrease of 20% in mortality relative to screening with chest X-rays³. Screening facilitates early detection of tumours, lowers mortality, and improves rates of surgical tumour removal⁴.

The benefits of screening are coupled with potential harms for both patients and the programmes themselves, such as radiation exposure, overdiagnosis, and false positives^3,5, not to mention accessibility concerns for those who fall outside eligibility parameters⁶. To improve the ratio of benefits to harms, refining the selection criteria for LCS participants is imperative. Given that with LDCT screening the number-needed-to-screen (NNS) is 320 screenings to prevent 1 lung cancer death³ and the number-needed-to-harm (NNH) is 59⁷, the incorporation of molecular biomarker research into screening protocols may prove crucial to improve their efficacy and reduce the rates of unnecessary invasive procedures.

A biomarker allows specific pathological diseases and processes to be identified earlier than the manifestation of symptoms, which may be observed through medical signs or a quantifiable substance. Enzymes, antibodies, receptors, as well as nucleic acids are biological examples of biomarkers⁸. Epithelial cells, bronchoalveolar lavage fluid, sputum, and exhaled breath constitute valuable reservoirs of molecular biomarkers. These sources are relevant to cancer screening as they provide valuable insights into the micro-environment of the tumour cells through non-invasive means.

Since biomarkers facilitate the identification of specific diseases and their underlying biological responses before the onset of clinical symptoms, their use in early detection cancer screening could complement existing methods. Their use extends through playing a role in determining eligibility in LCS programmes and in the management of indeterminate nodules⁹. Our research focus will be on non-invasive biomarkers used for LCS, exploring which biomarkers could form part of future screening programs and where future research will be best directed.

Existing literature

The literature on the use of biomarkers in lung cancer screening is growing, as this field strives to establish more robust and personalized screening programs. While individual biomarkers have shown promise, challenges remain in their application for routine clinical use. A recent review emphasized that while studies involving biomarkers for lung nodule triage have yielded encouraging results, no single biomarker is yet sufficient for widespread use. However, numerous new biomarker candidates have emerged, paving the way for future biomarker combinations¹⁰. Moreover, blood-based biomarkers, such as a 4-marker protein panel (Pro-SFTPB, CA125, CEA, and CYFRA 21-1), have demonstrated the potential to better individualize the risk of lung cancer and inform screening decisions. This panel, when combined with the PLCOm2012 risk prediction model, has shown improved identification of high-risk individuals compared to traditional screening criteria¹¹. Additionally, this study suggests that repeated biomarker measurements may improve sensitivity and early detection, further enhancing lung cancer risk prediction. These findings highlight the importance of ongoing research to refine biomarker strategies and explore their potential in clinical settings.

Knowledge gap, rationale and aim

This umbrella review recognises the potential role of biomarkers in addressing the limitations and drawbacks of traditional lung cancer screening methods. Biomarkers offer a promising avenue for improving risk stratification, which could result in more precise identification of individuals at higher risk of developing lung cancer, allowing for earlier diagnosis and commencement of treatment. This umbrella review aims to provide a comprehensive summary of the existing literature on the use of biomarkers in pre-diagnostic decision-making for lung cancer screening.

Objectives

We will complete this aim through the completion of the following specific objectives:

1. to systematically search the published literature and systematic review registries for relevant reviews;

2. to perform a bibliometric analysis of the identified systematic reviews and systematic review protocols (i.e., reviews which used a systematic search), categorising according to review type, publication status, publication year, country of the affiliated institution of the first and last author, the range of years over which the search will occur, and the number of eligible papers included (in the case of completed reviews);

3. to summarise the variation, between identified reviews, in their eligibility criteria relating to the population, intervention/exposure, comparison and outcome;

4. to illustrate the overlap of included primary studies, between completed systematic reviews, by a means appropriate to the number of primary studies (e.g. a network diagram);

5. to narratively summarise the findings of all identified systematic reviews of trials of biomarkers used pre-diagnostically in LCS, and establish whether a new or updated review is needed;

6. to produce a list of non-invasive biomarker tests which have been studied in a cohort of LCS patients, as well as the study methodology and event horizons studied, while acknowledging potential gaps in the evidence where these have been identified in the prior step;

7. to rank the putative performance of the biomarkers tests identified by this umbrella review by calculating the implied positive predictive value (PPV) and implied negative predictive value (NPV) for a range of empirically observed incidence values and event horizons.

Search strategy

Under three search themes, “non-invasive biomarkers”, “lung cancer”, and “screening”, a systematic search of literature will be conducted, with further manual search for other eligible articles through identified systematic reviews. The search strategy will include MeSH terms and keywords related to these themes. MEDLINE, Cochrane Database of Systematic Reviews, Embase, Scopus, and Web of Science are our databases of choice. A tailored search will be conducted to obtain the most relevant papers for our review. The following terms are included and outlined in Appendix 3: biomarker, lung cancer, screening. The approach to this search will be customized to meet the unique specifications of each database, using filters to locate primary studies as needed. The search will be limited to studies published in English from January 2010 to January 31, 2025. Full details of the search strategy will be outlined on the Open Science Framework.

To ensure the integrity of our dataset, we will initiate the data collection phase by employing the (Sr-Accelerator) Deduplicator tool to identify and eliminate any duplicate papers from our initial search outcomes.

Our current tailored search strategy focuses on a broad search. To convert our search terms to be applicable amongst each database, terms will be input into Systematic Review Accelerator (Sr-Accelerator) Polyglot Search tool and a result for each database will be used to retrieve search data.

Study selection

The systematic screening of imported literature will be streamlined using Rayyan, an AI-powered tool designed for efficient literature review management. We will apply the following exclusion criteria: no relevance to biomarker outcomes in LCS, non-human subjects, invasive biomarker/radiomic biomarker methods, and no controls. The selected literature will be directly relevant to the objectives of this umbrella review by strictly adhering to these criteria. The evaluation of titles and abstracts for inclusion in the review will be independently conducted by two reviewers (K.C.; C.S.). In instances of disagreement, a consensus will be sought through discussion. Should these discussions result in an impasse, a third reviewer (S.M.) is designated to arbitrate and make the final decision on the inclusion or exclusion of the paper.

Data collection

Following the deduplication process, the refined dataset will then be transferred into Rayyan for a screening and annotation phase. This phase will allow labels and notes to be made for all eligible papers post-screening, ensuring an efficient way to note data items. The steps taken to screen studies are summarised in Table 2.

The data extracted from review articles utilising a systematic search of the literature will be the review type, publication type, the range of years over which papers were extracted, the scope of the search in terms of the biomarker characteristics, other eligibility criteria, and the number of eligible papers included reviews. Data items to be extracted are outlined in Table 3.

Biomarker type. Biomarkers will be classified into serum tumour markers, exosomes, non-coding RNAs, micro-RNAs, long non-coding RNAs, circular RNAs, autoantibodies, metabolic biomarkers, cell-free DNA, DNA methylation, circulating tumour cells, volatile organic compounds, and combination studies. This classification is based on the recent work by Afridi et al.¹⁰. We will adapt this classification if biomarkers that do not fit these groups are found in the reviewed studies.

Outcomes. In relation to the biomarkers, the outcomes of interest will be:

Risk of bias assessment. The AMSTAR-2 tool¹⁵ will be used to assess the overall confidence in the results of each Systematic Review and Meta Analysis. Confidence in each will be assigned a rating of: high, moderate, low or critically low. The ROB-ME Tool¹⁶ for assessing risk of bias due to missing evidence in a meta-analysis will also be utilised.

Data synthesis

The data synthesis will follow a systematic approach, integrating the findings from identified systematic reviews to provide a comprehensive overview of biomarkers used in LCS. First, a bibliometric analysis will categorise the systematic reviews and protocols by review type, publication status, publication year, and country of the affiliated institution of the first and last authors. This will also include a summary of the range of years over which the search occurred and the number of eligible papers included in each review. Next, we will examine the variation in eligibility criteria across the reviews, focusing on differences in population, intervention/exposure, comparison, and outcome measures, which may impact the findings and generalisability of the results. To illustrate the overlap in primary studies included across the reviews, a network diagram will be used, highlighting the degree of redundancy in the evidence base.

The narrative synthesis will summarize the findings of the systematic reviews related to biomarkers used pre-diagnostically in LCS, assessing the need for a new or updated review in this area. A comprehensive list of non-invasive biomarker tests studied in LCS patient cohorts will be compiled, detailing study methodologies and the event horizons studied, while acknowledging gaps in the evidence identified through the variation analysis. Additionally, the performance of the identified biomarkers will be ranked by calculating their implied positive predictive value (PPV) and negative predictive value (NPV) across various empirically observed incidence values and event horizons. The synthesis will focus on drawing meaningful conclusions regarding the efficacy and utility of biomarkers in clinical practice, highlighting those with the greatest potential for future research and clinical integration.