Method

The protocol for this systematic review was prospectively registered in the PROSPERO database (CRD42022347826) and findings have been described in accordance with Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement.

Information sources and search strategy

Literature search strategies were developed by an information specialist and the first author (FJ and RO) using medical subject heading (MeSH) and text words related to ‘advanced adenoma’, ‘advanced serrated polyp’, ‘colorectal cancer’, ‘amino acids’, and ‘biomarker’. Articles were identified from MEDLINE, EMBASE and Cochrane Central Register of Controlled Trials on 29 November 2022. The full search is presented in Supplementary Table 1. The search was not limited to a specific language or date. Reference lists of included studies or relevant reviews identified through the search were scanned to check for any further eligible publications. An updated search was performed on 22-05-2025. To identify any significant new publications since the last update, a targeted screening of PubMed was conducted for the period from May to December 2025.

Selection process and patient population

The selection process was performed by two independent reviewers (RO and AH). Studies were included if they met the following criteria: (i) studies consisting of patients of 18 years or above with advanced adenoma, advanced serrated polyp or CRC; (ii) the diagnosis of AA, advanced serrated polyp or CRC was based on endoscopy combined with histopathology; (iii) studies on amino acid analysis for diagnostic biomarker discovery. The following definitions were used to classify advanced colorectal neoplasia: advanced adenomas were defined as conventional adenomas measuring ≥ 10 mm in diameter, and/or exhibiting villous histology (≥ 25% villous component), and/or high-grade dysplasia; advanced serrated polyps included sessile serrated lesions or hyperplastic polyps ≥ 10 mm, sessile serrated lesions with dysplasia or traditional serrated adenomas; and CRC was defined as histologically confirmed adenocarcinoma of the colon or rectum, irrespective of stage at diagnosis. A list of included amino acids is presented in Supplementary Table 2. Studies were excluded if they: involved patients with hereditary risk for colorectal neoplasia or patients with polyposis syndromes (e.g., Lynch syndrome, familial adenomatous polyposis, or serrated polyposis syndrome (SPS)); were non-original research or case reports; or lacked information on how controls were classified (e.g., absence of information on diagnostic work-up). Although the initial search strategy was broad and included general biomarker terms, we restricted the review during full-text screening to diagnostic applications to maintain a feasible and coherent scope for the review.

Data extraction and outcomes

From each study we extracted author, year, design, population, demographics, matrix, analytical method, amino acids measured, direction of change, effect size, p value, and histopathological data on advanced adenoma, advanced serrated polyps and CRC. Diagnostic performance (AUC, sensitivity, specificity) was extracted when reported for individual amino acids or panels consisting exclusively of predefined amino acids. If multiple time points or subgroups were reported, we prioritised baseline and predefined subgroup analyses. Data extraction was performed by RO using predefined forms and verified by AH to reduce errors. Missing or additional data were requested from study authors or retrieved from cited sources. Unresolved disagreements were resolved by a third author (SB). Results were synthesised using descriptive and tabular methods. Due to heterogeneity in study design, analytical methods, outcomes, and biospecimen types, no meta-analysis was performed. Findings were summarised narratively by matrix, lesion type, outcome type (e.g. concentration differences, diagnostic performance), and analytical approach (targeted or untargeted). To support interpretation, a forest plot was used to display AUCs, and a visual summary was generated to show the direction and consistency of amino acid changes across studies.

Quality assessment

The methodological quality of the included studies was assessed using the Newcastle–Ottawa Scale (NOS). For case–control studies, the original NOS tool was used, evaluating studies across three domains: selection, comparability, and exposure, with a maximum score of nine points. For cross-sectional studies, the modified version of the NOS as described by Modesti et al. (2016) was applied [13]. This adaptation, specifically developed for non-cohort observational designs, evaluates methodological quality across the domains of selection, comparability, and outcome, with a maximum score of ten points. Both versions are provided in the supplementary material page 3 and 4. Study assessment was performed by RO. In cases requiring clarification, a second author was consulted.

Statistics

If effect size as fold changes were not reported in the original publication, they were calculated manually based on the reported means or medians. For each study, it was clearly indicated whether fold changes were extracted or calculated in the corresponding table. For the study by Bosch et al., raw data were publicly available and used to perform additional statistical analyses [14]. Mann–Whitney U tests were conducted to compare CRC versus controls, advanced adenomas versus controls, and CRC plus advanced adenomas versus controls using RStudio (version 4.4.3).