Advanced imaging techniques at the crossroads of cholangiocarcinoma and liver transplantation: Can we bridge the gap?

Abstract

In order to evaluate emerging imaging strategies for optimizing cholangiocarcinoma (CCA) assessment in liver transplantation (LT) candidates, addressing gaps in standardization, diagnostic ambiguity, and equitable access. Critical analysis of current evidence and innovations in CCA imaging, focusing on three pillars: (1) Adaptation of Liver Imaging Reporting and Data System for standardized reporting; (2) Integration of artificial intelligence (AI)-driven radiomics for risk stratification; and (3) Expanded utilization of contrast-enhanced ultrasound (CEUS) in resource-limited settings. Current imaging criteria for LT eligibility in CCA rely heavily on tumor size and vascular invasion, but lack standardized protocols for lesion characterization in cirrhotic livers. Liver Imaging Reporting and Data System, validated for hepatocellular carcinoma, shows promise in reducing interobserver variability when adapted to CCA-specific features (e.g., targetoid appearance on magnetic resonance imaging). AI-driven radiomics can predict microvascular invasion and post-LT recurrence risk with 85% accuracy in preliminary studies, while CEUS demonstrates 92% specificity for differentiating intrahepatic CCA from dysplastic nodules in cirrhosis. A harmonized approach combining standardized reporting systems, AI-powered analytics, and accessible imaging modalities like CEUS could redefine LT pathways for CCA. Collaborative efforts between radiologists and transplant teams are essential to translate these innovations into equitable, precision-driven care.

Key Words: Cholangiocarcinoma; Liver transplantation; Artificial intelligence; Precision medicine; Cancer

Core Tip: This perspective advocates for paradigm shifts in cholangiocarcinoma imaging for transplantation: Liver Imaging Reporting and Data System standardization to reduce diagnostic variability; artificial intelligence radiomics for recurrence risk prediction; and contrast-enhanced ultrasound adoption to bridge resource gaps. Together, these innovations can transform ambiguous evaluations into precision-driven pathways, optimizing post-transplant survival.

INTRODUCTION

Cholangiocarcinoma (CCA) represents a critical frontier in liver transplantation (LT), where 5-year survival remains suboptimal (50%-65%) despite stringent selection protocols[1]. which involve complex, staged evaluations of both recipients and potential living donors[2]. Current radiological criteria centered on tumor size and vascular invasion fail to resolve key challenges: Differentiating malignancy in cirrhotic livers[3], predicting post-transplant recurrence[4], and ensuring equitable access across resource settings[5]. Zhou et al[6] recent review underscores radiology’s pivotal role, yet rapid technological advancements now demand a reimagined approach. This article envisions a future where advanced imaging transcends descriptive reporting to become the cornerstone of precision transplantation, contributing to a more robust prediction of patient prognosis[7,8].

CURRENT PROBLEMS: AMBIGUITY IN CIRRHOTIC LANDSCAPES

The cirrhotic liver presents a unique diagnostic battleground. Intrahepatic CCA frequently mimics dysplastic nodules on conventional imaging, with overlapping features like arterial hyperenhancement leading to false negatives[9]. This ambiguity delays transplant eligibility decisions and risks excluding curable patients[10]. Compounding this, LT candidates often exhibit heterogeneous cirrhosis etiologies, further complicating lesion characterization[11]. The interobserver variability quagmire subjective interpretation of imaging findings fuels inconsistent LT eligibility assessments. A 2023 multicenter study revealed only 68% inter-radiologist agreement in classifying indeterminate liver lesions as CCA[12]. Such variability stems from no standardized reporting and undermines objectivity in life-altering transplant decisions[13]. Noteworthy, the socioeconomic disparities in magnetic resonance imaging/computed homography access create inequitable pathways to transplantation[14]. In low-resource regions, diagnostic delays exceeding 8 weeks reduce transplant eligibility by 40%[15], highlighting an urgent need for portable alternatives[16].

In this context, the Liver Imaging Reporting and Data System (LI-RADS), validated for hepatocellular carcinoma[17], offers a template for standardizing CCA reporting. Adapting LI-RADS criteria to incorporate CCA hallmarks, such as peripheral rim enhancement or targetoid diffusion restriction, could elevate diagnostic specificity for CCA-containing tumors up to 100%[18,19]. Quantitative apparent diffusion coefficient thresholds may further reduce ambiguity in cirrhotic nodules, though multicenter validation remains essential[20,21]. In the same line, deep learning algorithms extract subvisual features to predict microvascular invasionand stratify post-LT recurrence risk[22,23], a concept that has also shown promise in predicting recurrence after locoregional bridge therapies[24]. Ji et al[25] developed two radiomics-based models integrating pre-operative and post-operative variables, which achieved high predictive performance with lower prediction error for tumor recurrence, allowing prediction of individual risk and distinct profiles of recurrent tumor.

Federated learning approaches now enable model training across institutions without data sharing[26], addressing privacy concerns. Additionally, a pioneer predictive model developed to differentiate intrahepatic cholangiocarcinoma (ICC) from hepatocellular carcinoma in high risk patients report that the most useful contrast-enhanced ultrasound (CEUS) features for ICC were rim enhancement, early washout, intratumorally vein, obscure boundary of intratumorally non-enhanced area, and marked washout, significantly improving the differentiation from ICC and hepatocellular carcinoma in ≤ 5.0 cm size tumors[27]. Notably, CEUS portability reduces diagnostic costs up to 60% compared to magnetic resonance imaging[28], enabling rapid triage in resource-limited settings[29] (Table 1).

Table 1 Emerging imaging strategies for cholangiocarcinoma assessment in liver transplant candidates.

Imaging strategy	Key features	Strengths	Limitations	Ref.
Adapted LI-RADS for CCA	Targetoid appearance, rim enhancement, ADC thresholds	Standardization; improves interobserver agreement	Requires validation; originally developed for HCC	Hong et al[12]; Majeed et al[18]; Jiang et al[19]; Pankaj Jain et al[20]
AI-driven radiomics	Texture analysis; deep learning for microvascular invasion, recurrence risk	High predictive accuracy; individualized risk stratification	Limited generalizability; needs multicenter training	Ahmadzadeh et al[3]; Zerunian et al[22]; Barcena-Varela et al[23]; Lindner et al[24]; Ji et al[25]
Contrast-enhanced ultrasound	Rim enhancement, early washout, intratumorally vein	Portable; low cost; high specificity in cirrhosis	Operator-dependent; limited spatial resolution	O’Brien et al[16]; Wu et al[27]; Smajerova et al[28]

LI-RADS: Liver Imaging Reporting and Data System; CCA: Cholangiocarcinoma; AI: Artificial intelligence; ADC: Apparent diffusion coefficient; HCC: Hepatocellular carcinoma.

CONCLUSION

Prospective trials must link LI-RADS adaptations and radiomic biomarkers to post-LT survival endpoints[10]. Initiatives like the International LT Society-Radiology Collaborative could harmonize protocols[30]. In this line, national societies should advocate for CEUS inclusion in LT guidelines and fund training programs in underserved regions. By uniting LI-RADS standardization, radiomic prognostication, and CEUS-driven accessibility, we can replace diagnostic ambiguity with precision. This triad transcends technological novelty and embodies a commitment to equitable, survival-optimized transplantation. Nonetheless, translating these innovations into clinical practice is not without challenges. Variability in regulatory approval for CEUS, the need for standardized training across institutions, and disparities in technological infrastructure may limit immediate adoption. Furthermore, integrating LI-RADS adaptations and radiomic biomarkers into transplant decision-making requires prospective validation and consensus among multidisciplinary teams. Recognizing these barriers is essential, as addressing them through collaborative frameworks will ultimately determine the feasibility and sustainability of these advances in routine transplantation care.