Predictors of hepatocellular carcinoma in LR-M category lesions, a multi-institutional analysis
Bibliographic record
Abstract
The Liver Imaging Reporting and Data System (LI-RADS, LR) provides a framework for diagnosing hepatocellular carcinoma (HCC, LR-5). However, not all HCCs meet LR-5 criteria and are instead categorized as LR-M, probably or definitely malignant but not specific for HCC, necessitating biopsy for diagnosis. The purpose is to identify factors associated with HCC in LR-M observations. This is an IRB-approved, retrospective analysis of participants from 8 institutions that had a LR-M observation on CT or MRI with corresponding histopathologic diagnosis. Demographics and biochemical data were examined. Central review using the LI-RADS v2018 algorithm was performed. Kappa statistics defined inter-reader agreement. Random forest and logistic regression analyses generated a model for HCC diagnosis. 162 participants with 162 LR-M observations were included. 46% of observations (74/162) were HCC and 37% were cholangiocarcinoma (60/162). Two of 34 imaging features– observation size and intra-lesion iron– showed moderate to strong inter-reader agreement (Kappa ≥ 0.60) while the remainder showed weak or no agreement (Kappa < 0.60). Random forest analysis showed biochemical features to be more predictive of HCC than imaging features. Logistic regression analysis of a model based on INR and AFP provided a 72% sensitivity and 61% specificity for HCC by Youden’s index and a 90% specificity threshold yielded 38% sensitivity, 75% positive predictive value, and 66% negative predictive value. Our results show INR and AFP are associated with HCC in LR-M observations. A high-specificity threshold may assist in the non-invasive diagnosis of HCC in the appropriate setting. In certain at-risk patients with a LR-M observation on diagnostic imaging, serum AFP and INR maybe useful tools for the non-invasive diagnosis of HCC. • In this cohort of 162 LR-M observations, the most common diagnoses were HCC (46%, 74/162) or cholangiocarcinoma (37%, 60/162). • Serum INR and AFP were associated with HCC in LR-M observations and no imaging features were predictive of HCC. • While the imaging-based diagnosis of HCC in LR-M observations remains a challenge, a high specificity model based on INR and AFP may suggest the diagnosis; however, this threshold model comes at the expense of sensitivity.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".