Liquid-based cytology in the era of multi-omics and artificial intelligence integration

Dear Editor,

Liquid-based cytology (LBC) is a cytopreparatory technique that involves collecting exfoliated or aspirated cells in a liquid fixative medium. It was developed in the 1990s as an alternative to conventional Pap smears for cervical cancer screening.^[1] Unlike traditional smears, where cells are directly transferred to a slide, LBC suspends collected cells in a preservative solution, which is then processed to remove debris and produce a uniform monolayer of cells. This technique significantly improves sample adequacy, clarity, and overall diagnostic quality.

Initially used in gynecologic cytology, LBC has now largely replaced conventional smears in many countries due to its superior specimen quality and the ability to perform ancillary tests such as human papillomavirus (HPV) deoxyribonucleic acid (DNA) detection from the same sample. Beyond cervical cytology, LBC has proven valuable in non-gynecologic settings, including urine, thyroid, respiratory tract samples (e.g., bronchoalveolar lavage), body fluids, and fine-needle aspiration (FNA) biopsies of various organs. The method facilitates superior cell preservation, cleaner backgrounds, and compatibility with immunocytochemistry and molecular testing.^[1] Originally developed to overcome the limitations of conventional cytology smears, LBC offered improved sample adequacy, cleaner backgrounds, and easier storage of cellular material.^[2] Over time, LBC has evolved far beyond its initial role. It is now increasingly recognized as a powerful, versatile biospecimen that supports a wide spectrum of advanced molecular diagnostics. As precision medicine continues to expand, LBC is gaining new relevance as a foundational platform for integrated testing – spanning genomics, transcriptomics, proteomics, and metabolomics – and for incorporating artificial intelligence (AI) into diagnostic workflows.^[3] These innovations are reshaping the identity of cytopathology, placing it at the cutting edge of personalized medicine.

Recent research has expanded the scope of LBC into liquid biopsy-based cancer diagnostics, including analysis of circulating tumor cells and cell-free DNA, particularly for early detection and monitoring of malignancies. Advances in digital cytopathology and automated image analysis of LBC slides are improving diagnostic accuracy and interobserver consistency. Furthermore, LBC is being studied for its ability to improve diagnostic yield in low-cellularity specimens, and its use is being evaluated across various screening and triage protocols in clinical settings.^[1,4]

Central to this transformation is the ability of LBC to maintain both cellular morphology and molecular integrity. The fluid-based collection and processing of cytologic specimens ensure excellent preservation of nucleic acids and proteins without compromising the morphological integrity of cells. This dual preservation is crucial for performing multi-modal analyses – such as combining microscopic examination with high-throughput molecular techniques – from a single cytology sample [Figure 1].^[4] In genomic testing, LBC-derived DNA and ribonucleic acid have proven suitable for polymerase chain reaction (PCR), quantitative PCR, and next-generation sequencing (NGS). These techniques enable the detection of key mutations such as EGFR and ALK in lung cancer, BRAF in thyroid nodules, and various HPV genotypes in cervical cytology.^[5] Multiple studies have demonstrated the feasibility of NGS from LBC samples obtained through FNA or brushings, further expanding its clinical applicability.^[6] More recently, transcriptomic applications using LBC samples have provided gene expression profiles that assist in differentiating benign from malignant lesions and predicting therapeutic responses.^[6]

Close

Figure 1:

Evolving role of liquid-based cytology in precision medicine through integration with multi-omics and artificial intelligence (AI). The schematic diagram illustrates how conventional liquid-based cytology preparations, traditionally used for morphological assessment, are now being repurposed for downstream applications in genomics, transcriptomics, proteomics, metabolomics, and radiomics. AI and machine learning algorithms enhance diagnostic precision by enabling digital cytopathology, image-based classification, and predictive analytics. This integration supports biomarker discovery, tumor profiling, and personalized therapeutic strategies, thereby redefining cytology from a purely diagnostic tool to a central platform in precision oncology and systems medicine. Image generated through Microsoft PowerPoint (Microsoft Corporation [Redmond, Washington, USA]; version PowerPoint 2021) and Microsoft Paint (Microsoft Corporation [Redmond, Washington, USA], included with Windows operating systems, with version Windows 10).

Export to PPT

Beyond genomics, proteomic analysis of LBC specimens is increasingly feasible, where protein extraction from preserved cells supports biomarker discovery and validation. For instance, assessment of programmed death-ligand 1 (PD-L1) expression in lung cancer is now frequently performed on cytologic samples obtained through minimally invasive procedures such as endobronchial ultrasound-guided FNA.^[7] Similarly, emerging evidence suggests that metabolomic profiling of LBC-derived fluids – including pleural and peritoneal effusions – can identify distinct metabolic signatures indicative of malignancy or therapeutic outcomes.^[8] These cumulative applications are driving broader integration of LBC into multi-omics workflows, transforming how cytologic samples are utilized in both clinical and research contexts.^[8] Such multi-omics integration enables a systems-level understanding of disease processes and patient stratification.

Despite these advancements, several challenges remain in implementing multi-omics analyses using LBC. The quantity and quality of cellular material may limit the feasibility of multiple downstream assays. Although LBC improves cell preservation compared to conventional smears, some clinical samples still present low cellularity.^[9] In addition, the pre-analytical phase – including sample fixation, storage, and processing – can significantly influence molecular test performance.^[10] Standardization across institutions remains a critical goal, as protocol variations can introduce inconsistencies that affect reproducibility. Furthermore, the interpretation and integration of complex multi-omics data require advanced bioinformatic tools and interdisciplinary collaboration involving cytopathologists, molecular biologists, and computational scientists.^[11]

Alongside molecular advancements, AI is emerging as a transformative force in cytopathology. AI-powered diagnostic tools – particularly those using deep learning algorithms – are being developed to assist in image analysis, pattern recognition, and classification of cytologic specimens.^[12] In cervical cancer screening, AI has demonstrated significant potential in automating the detection of atypical squamous cells and reducing interobserver variability.^[13] With the rise of digital pathology and whole-slide imaging, AI systems can now be trained on large datasets to detect subtle cytomorphologic features, offering highly accurate diagnostic support that complements the expertise of human cytopathologists.^[13]

An especially exciting frontier is the convergence of AI with multi-omics data derived from LBC specimens. By combining cellular morphology with molecular profiles from the same sample, AI models can be trained to predict diagnostic categories as well as underlying genetic alterations, prognostic markers, and therapeutic responses.^[14] In lung cancer cytology, for example, AI could integrate image-based features with EGFR mutation status and PD-L1 expression to stratify patients for targeted therapy.^[15] This fusion of computational pathology and molecular diagnostics represents a paradigm shift, enabling a systems-level understanding of disease that surpasses the capabilities of morphology or molecular data alone.^[15]

Clinical implementation of this integrated approach is already taking shape. In cervical cancer screening, LBC samples are used for concurrent HPV DNA testing, methylation marker detection, and AI-assisted morphologic triage – enhancing risk stratification and follow-up precision.^[16] In thyroid cytology, particularly indeterminate cases (Bethesda Category III), LBC-based molecular panels and AI-assisted assessment are helping reduce unnecessary surgeries by improving diagnostic certainty.^[17] Likewise, in thoracic oncology, LBC specimens from minimally invasive procedures are being leveraged for multiplexed molecular testing – including NGS for driver mutations and immunohistochemical analysis of immune checkpoint markers.^[16]

To fully realize the potential of LBC in this evolving landscape, several critical steps must be taken. First, standardization of laboratory protocols – including cell fixation, nucleic acid extraction, and protein isolation – is essential to ensure consistent sample quality. Second, interdisciplinary collaboration is key. Cytopathologists must partner with molecular pathologists, bioinformaticians, and AI developers to design diagnostic workflows that maximize informational yield.^[18] Third, infrastructure must be enhanced to support digital and AI adoption, requiring investments in imaging systems, computing hardware, and secure data storage. Finally, regulatory frameworks must evolve to address transparency, data privacy, and equitable access to advanced diagnostic tools.^[19]

LBC is rapidly emerging as a central platform in the era of multi-omics and AI. Its capacity to support both morphologic evaluation and molecular testing positions it uniquely within the expanding field of precision diagnostics. As cytology continues to move beyond its traditional boundaries, it is poised to play a vital role in guiding personalized therapy, monitoring disease progression, and discovering new biomarkers. The integration of LBC with multi-omics and AI is not just enhancing diagnostic capabilities – it is redefining the future of cytopathology.