Magnetic resonance imaging R2* mapping for evaluating Ki-67 proliferation index and cytokeratin 19 expression in solitary iron-sparing hepatocellular carcinoma: a retrospective single-center study

Introduction

Hepatocellular carcinoma (HCC) is the most common primary liver cancer and the third leading cause of cancer-related death worldwide (1). Although advances in surgery and imaging have improved prognosis, postoperative recurrence remains high (2-4). In addition, marked tumor heterogeneity contributes to substantial variability in prognosis even among patients receiving similar treatment. Reliable biomarkers of tumor aggressiveness are therefore needed for risk stratification among patients with HCC. Ki-67, a nuclear marker of proliferation, reflects tumor growth activity (5) and is associated with higher malignant potential, shorter recurrence-free survival (RFS), and worse overall survival (OS) in patients with HCC (6-8). Cytokeratin 19 (CK19), a marker of cholangiocytes and hepatic progenitor cells (9), is linked to aggressive tumor behaviors—including capsule disruption, microvascular invasion, vascular tumor thrombus, and nodal metastasis—resulting in earlier recurrence and poor survival (10-14). Thus, evaluating the expression of Ki-67 and CK19 is critical to accurately guiding risk stratification, individualized management, and prognostic assessment among patients with HCC.

The diagnosis of Ki-67 and CK19 is conducted mainly via preoperative biopsy or postoperative pathology (15), both of which are limited by procedural risk and sampling error (16). Preoperative noninvasive imaging, especially magnetic resonance imaging (MRI), offers a potential alternative (14,17-23). Certain MRI features can predict Ki-67 and CK19 expression (14,17-20), and advanced methods such as deep learning, texture analysis, and radiomics have demonstrated predictive capability (21-23). However, the diagnostic efficacy of MRI is influenced by imaging features and analysis methods, and its value in predicting Ki-67 and CK19 expression remains controversial.

Iron is a key trace element for maintaining cellular function (24) and may also be involved in the occurrence and development of HCC (25-27). According to one study (26), patients with hereditary hemochromatosis have a significantly increased risk of developing HCC, approximately 200 times that of the general population. Meanwhile, the growth of tumors can be inhibited by depriving cancer cells of essential iron (25,27). The tumor lesions of patients with HCC share a common feature: they lack iron deposition, even when formed in iron-loaded livers. The 2018 version of the Liver Imaging Reporting and Data System (LI-RADS v2018) lists intratumoral iron‑sparing as an ancillary feature favoring malignancy, but this is not HCCspecific (28). However, the relationship between hepatic and intratumoral iron content and tumor-related markers (e.g., Ki-67 and CK19) in ironsparing HCC remains unclear. R2* relaxometry is the preferred noninvasive method for iron quantification (29). Only one study has examined the relationship between hepatic and intratumoral R2* values and the Ki-67 proliferation index (PI) in HCC (30). Therefore, research using preoperative hepatic and intratumoral R2* values to assess Ki-67 and CK19 is sparse, and the relationship between these factors has not been conclusively determined.

Iron-sparing HCC may provide a clinically relevant setting in which tumor iron metabolism and background hepatic iron status can be assessed simultaneously. Previous research in this area (14,17-23) has mainly evaluated conventional MRI features, radiomics, or serum alpha-fetoprotein (AFP) for Ki-67 and CK19 prediction, whereas quantitative R2* mapping directly reflects tissue iron-related relaxation (29) and may capture a different biological dimension (25-27). However, available evidence regarding hepatic or intratumoral R2* values in relation to Ki-67 and CK19 is sparse, and the relative contributions of background hepatic R2* and intratumoral R2* values have not been clarified in a surgically confirmed cohort of patients with iron-sparing HCC.

Therefore, this study aimed to determine whether preoperative hepatic and intratumoral R2* values are associated with Ki-67 PI and CK19 expression in patients with iron‑sparing HCC. We further assessed whether these quantitative MRI parameters, both alone or combined with established clinical indicators, could support the noninvasive preoperative evaluation of tumor aggressiveness. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0434/rc).

Methods

The study was reported according to the principles of retrospective clinical imaging research, and all analyses were based on data available before surgery and on postoperative histopathological reference standards. This retrospective single-center study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and was approved by the Institutional Review Board of The Third Affiliated Hospital of Sun Yat-sen University (approval No. II2025-229-01). The requirement for informed consent was waived due to the retrospective nature of the analysis.

Patients

We enrolled consecutive patients with pathologically confirmed HCC who underwent resection between May 2018 and May 2025 and had preoperative 3.0-T MRI with R2* maps showing intratumoral iron-sparing. Intratumoral iron-sparing was defined as a tumor signal lower than the background liver (31) on R2* maps. The other inclusion criteria were (I) age ≥18 years; (II) surgical resection; (III) solitary HCC; (IV) complete postoperative Ki-67 and CK19 data; and (V) MRI scanning performed within 1 month before surgery. Meanwhile, the exclusion criteria were (I) previous antitumor therapy; (II) a tumor diameter <10 mm; (III) significant R2* map artifacts; (IV) intratumoral hemorrhage; and (V) tumor steatosis. According to Hong et al.’s method (32), a proton density fat fraction (PDFF) <2.2% indicated noise, and lesions exceeding this threshold were considered to be steatotic and excluded from the analysis.

MRI examination protocol

Imaging was performed with 3.0-T scanners (Discovery MR750 and SIGNA Architect; GE HealthCare, Chicago, IL, USA). Patients fasted for 6–8 hours and received breath-hold training. Contrast agent (0.1 mmol/kg) was administered via a dual-head power injector (Spectris Solaris EP; Medrad, Bayern, Leverkusen, Germany) and included extracellular agents (n=3; Magnevist, Bayer), gadobenate dimeglumine (n=57; MultiHance, Bracco, Milan, Italy), or gadoxetate disodium (n=20; Primovist, Bayer), which was followed by a 20-mL saline flush. The protocol included T2-weighted imaging, diffusion-weighted imaging, chemical shift imaging, liver acquisition with volume acceleration dynamic contrast-enhanced imaging (with the hepatobiliary phase at 2 h for MultiHance and at 20 min for Primovist), and an axial six-echo three-dimensional gradient-echo sequence (mDixon) for PDFF and R2* mapping.

R2* maps were acquired in a single breath-hold. The SIGNA Architect parameters were as follows: first echo time (TE1) =0.8 ms, repetition time (TR) =6.2 ms, echo time interval (ΔTE) =1.2 ms, flip angle =3°, number of echoes =6, echo train length =3, matrix =128×128, field of view (FOV) =40 cm × 40 cm, number of slices =36, slice thickness =6 mm, voxel size =3.1×3.1×6.0 mm³, and bandwidth =111.11 kHz. The Discovery MR750 parameters were as follows: TE1 =0.8 ms, TR =5.4 ms, ΔTE =1.2 ms, flip angle =3°, number of echoes =6, matrix =128×128, FOV =36 cm × 36 cm, number of slices =24, thickness =8 mm, and bandwidth =125.00 kHz. This sequence could automatically generate R2* maps without the need for post-processing.

MRI analysis

Two radiologists (S.C. and C.D. with 8 and 7 years of abdominal MRI experience, respectively) who were blinded to the clinical data, Ki-67 status, and CK19 status but aware of HCC diagnosis and tumor location independently analyzed the images. On R2* maps, they manually drew regions of interest (ROIs) for the tumor (on the largest-diameter slice, maximized for size) and liver parenchyma (two ROIs in the right lobe and one in the left lobe, with vessels, ducts, and lesions being avoided). Each performed three measurements per tumor, which were averaged. The mean hepatic R2* value was calculated from the three parenchymal ROIs. Both radiologists performed measurements for assessment of agreement, but data from the more experienced one were used for analysis. Hepatic R2* was analyzed both as a continuous variable in group comparisons and as a categorical variable, with 136 s⁻¹ serving as the threshold, in line with other multicenter research on liver iron quantification (33).

Background liver cirrhosis was assessed on contrast-enhanced MRI. The LI-RADS v2018 criteria (34) were used to measure tumor size (mm) and evaluate major and ancillary features, tumor in vein (TIV), LR-M (probably or definitely malignant, not specific for HCC) features, and category.

Clinical data

Data collected from the hospital picture archiving and communication system included age, sex, etiology, Child-Pugh class, and serological parameters within 2 weeks before surgery, including AFP (ng/mL), total bilirubin (TBIL; µmol/L), alanine aminotransferase (ALT; U/L), aspartate transaminase (AST; U/L), albumin (ALB; g/L), platelet (PLT) count (10⁹/L), and international normalized ratio (INR). AFP was analyzed continuously in group comparisons and categorically at 400 ng/mL because this threshold is widely used in HCC risk stratification and prognostic assessment (4,35).

Histopathologic analysis

The pathological data were obtained by one author (J.X.) who had no knowledge of clinical or imaging information. Pathology data included diagnosis, location, differentiation grade, presence of hemorrhage and necrosis, and expression of Ki-67 and CK19. Ki-67 and CK19 expression was assessed via standard immunohistochemistry. Ki-67 positivity (brown-yellow nuclei) was recorded as the PI (0–100%). According to the cohort’s median PI, lesions were categorized as having a low (≤20%) or high (>20%) Ki-67 PI. CK19 positivity was defined as membranous/cytoplasmic staining in ≥5% of tumor cells (9). Lesions were matched by size and location for imaging-pathology correlation.

Missing data, measurement comparability, and control for bias

Only patients with complete preoperative MRI, clinical information required for the analyses, and postoperative Ki-67 and CK19 results were included in the study. Therefore, no imputation, last-observation-carried-forward procedure, or other missing-data replacement method was applied. Measurement comparability across the Ki-67 and CK19 groups was ensured by the use of the same MRI platforms, acquisition protocols, ROI rules, laboratory time window, and histopathological reference standard for all included patients. Selection bias was addressed through the implementation of consecutive enrollment and use of predefined inclusion and exclusion criteria, information bias was reduced via blinded imaging review and blinded pathological data extraction, measurement bias was evaluated according to the interreader intraclass correlation coefficient (ICC), and confounding bias was addressed via multivariate exact logistic regression with parsimonious model construction.

Sample size considerations

Because the study involved the retrospective analysis of a specific imaging phenotype, the sample size was determined by the number of consecutive eligible patients enrolled during the study period rather than by prospective recruitment. All patients meeting the predefined eligibility criteria were included. The final sample comprised 80 patients, with 43 high-Ki-67 PI events and 16 CK19-positive events. Exact logistic regression was used because of the modest sample size, low event counts for positive CK19 expression, and low-frequency categorical predictors. To avoid overfitting in the multivariate model, only variables selected from univariate analysis and with clinical relevance were considered.

Statistical analysis

Statistical analysis was performed with SPSS version 20.0 (IBM Corp., Armonk, NY, USA), R version 4.5.1 (The R Foundation for Statistical Computing, Vienna, Austria), and MedCalc software (MedCalc Software, Ostend, Belgium). The Shapiro-Wilk test and visual inspection of distributions were used to assess normality. Continuous variables are expressed as the mean ± standard deviation or as the median and IQR, as appropriate, and comparisons between two independent groups were made via the t-test or Mann-Whitney U test. Categorical variables are presented as frequency (percentage) and were compared with the Pearson Chi-squared test or the Fisher exact test. Outliers were screened with distribution plots and source-record verification; no observation was excluded solely due to an extreme but clinically plausible value. Nonnormal quantitative variables were handled with nonparametric tests rather than with distributional transformation.

Exact logistic regression analysis was performed to identify predictors of a high Ki-67 PI and positive CK19 expression. Variables with P <0.1 in the univariate analysis were included in multivariate analysis. Multivariate models were constructed using a backward stepwise strategy while limiting the number of retained predictors according to the available event counts. Potential confounding was addressed by entering eligible clinical and imaging variables into the corresponding multivariate model; variables representing closely related R2* constructs were assessed carefully to avoid overparameterization. No subgroup analysis or interaction-effect analysis was prespecified or performed. The area under the receiver operating characteristic curve (AUC) was calculated to evaluate the diagnostic performance of the prediction models.

Interreader agreement for R2* measurements in HCC lesions and liver parenchyma was assessed according to the ICC. Agreement was categorized as poor (ICC <0.50), moderate (0.50≤ ICC <0.75), good (ICC =0.75–0.90), or excellent (ICC >0.90).

All tests were two-sided, and P<0.05 was considered statistically significant.

Results

Patient characteristics

A total of 80 patients with solitary iron‑sparing HCC were included in the study (mean age 55.11±9.69 years), including 75 (93.75%) males. The flowchart of patient inclusion is presented in Figure 1, and patients’ baseline characteristics are summarized in Table 1. All included patients had complete MRI R2* measurements, clinical variables used in the analyses, and postoperative Ki-67 and CK19 immunohistochemical results. Regarding etiology, 69 (86.25%) HCC cases were attributed to hepatitis B virus (HBV), 2 (2.50%) to hepatitis C virus (HCV), 5 (6.25%) to anemia (unknown etiology), and 4 (5.00%) to unknown causes. In terms of Child-Pugh class, 78 (97.50%) patients were class A and 2 (2.50%) were class B. The median AFP level was 11.78 [interquartile range (IQR), 4.00–165.51] ng/mL.

Figure 1 Flowchart of patient selection. CK19, cytokeratin 19; HCC, hepatocellular carcinoma; MRI, magnetic resonance imaging; PI, proliferation index.

The median interval between preoperative MRI and surgery was 7 (range, 1–24) days. Of the 80 patients included, 47 (58.75%) had cirrhosis, while 33 (41.25%) did not. No statistically significant difference was observed in the median hepatic R2* values between the cirrhosis (152.73 s⁻¹, IQR, 99.13–215.25 s⁻¹) and non-cirrhosis (128.33 s⁻¹, IQR, 94.39–179.19 s⁻¹) groups (P=0.247). Among the 69 patients with HBV infection, 35 received antiviral therapy while 34 did not. There was no statistically significant difference in the median hepatic R2* values between the antiviral treatment group (152.73 s⁻¹, IQR 99.07–193.60 s⁻¹) and the non-treatment group (124.59 s⁻¹, IQR, 98.92–210.50 s⁻¹) (P=0.885). The median tumor size was 32.04 (IQR, 21.17–44.68) mm. The median hepatic R2* value was 128.68 (IQR, 99.07–196.30) s⁻¹, the median intratumoral R2* value was 34.95 (IQR, 21.67–49.93) s⁻¹, and the median difference between the two was 91.47 (IQR, 67.89–147.13) s⁻¹. The distribution of LI‑RADS category was as follows: LR‑4, 11 (13.75%); LR‑5, 62 (77.50%); LR‑M, 5 (6.25%); and LR‑TIV, 2 (2.50%).

Pathological analysis revealed that among the included cases, 10 were poorly differentiated, 69 moderately differentiated, and 1 well-differentiated. Additionally, 5 had HCC with hemorrhage, and 19 had HCC with necrosis. No statistically significant difference was observed in intratumoral R2* values across the differentiation groups (P=0.248). Similarly, there were no significant differences in R2* values between HCC lesions with and without hemorrhage (P=0.787) or between those with and without necrosis (P=0.363).

Clinical and imaging features based on the Ki-67 PI

Among the 80 patients, 37 (46.25%) exhibited a low Ki-67 PI, while 43 (53.75%) had a high Ki-67 PI. In the low-Ki-67 PI group, the mean age was 56.46±7.85 years, with 35 (94.59%) males and a median tumor size of 30.38 (IQR, 20.74–45.22) mm. In the high-Ki-67 PI group, the mean age was 53.95±10.98 years, with 40 (93.02%) males and a median tumor size of 32.82 (IQR, 22.12–43.75) mm. The high Ki-67 PI group, as compared to the low Ki-67 PI group, showed significantly higher AFP levels [101.80 (IQR, 2.84–691.30) vs. 6.58 (IQR, 4.09–26.52) ng/mL; P=0.035], significantly lower hepatic R2* values [112.90 (IQR, 79.63–185.65) vs. 162.67 (IQR, 106.33–209.78) s⁻¹; P=0.038], and a significantly smaller difference between hepatic and intratumoral R2* values [81.90 (IQR, 55.33–120.10) vs. 113.20 (IQR, 80.46–165.14) s⁻¹; P=0.022]. No significant differences were observed in other clinical or imaging characteristics between the two groups (P>0.05). The details of the Ki-67 PI results are shown in Table 2.

Clinical and imaging features based on CK19 expression

Among the 80 patients, 64 (80%) had negative CK19 expression, while 16 (20%) had positive expression. In the CK19-negative group, the mean age was 55.34±9.63 years, with 60 (93.75%) males and a median tumor size of 38.83 (IQR, 22.18–45.32) mm. In the CK19-positive group, the mean age was 54.19±10.19 years, with 15 (93.75%) males and a median tumor size of 28.10 (IQR, 16.91–38.20) mm. The CK19-positive group, as compared to the CK19-negative group, had significantly higher AFP levels [557.62 (IQR, 2.49–937.73) vs. 10.23 (IQR, 4.07–113.87) ng/mL; P=0.047] and a significantly smaller difference between hepatic and intratumoral R2* values [83.89 (IQR, 50.32–92.73) vs. 109.07 (IQR, 71.39–161.28) s⁻¹; P=0.050]. No other significant differences were observed in clinical or imaging characteristics between the two groups (P>0.05). Details of the CK19 expression results are shown in Table 2.

Predictors of high Ki-67 PI

Univariate exact logistic regression analysis revealed that the factors associated with high Ki-67 PI were AFP >400 ng/mL [odds ratio (OR) =14.60, 95% confidence interval (CI): 3.32–138.12; P<0.001), hepatic R2* value ≥136 s⁻¹ (OR =0.37; 95% CI: 0.15–0.89; P=0.027), and the difference between hepatic and intratumoral R2* values (OR =0.99; 95% CI: 0.99–1.00; P=0.039). Multivariate analysis confirmed that the independent predictors of a high Ki-67 PI were AFP >400 ng/mL (OR =19.04; 95% CI: 2.98–121.46; P=0.002) and a hepatic R2* value ≥136 s⁻¹ (OR =0.27; 95% CI: 0.10–0.75; P=0.012). The logistic regression model can be expressed as follows: Logit P = 0.32 + 2.95 × AFP – 1.32 × R2*. The combined predictive model achieved an AUC of 0.771 (95% CI: 0.664–0.858) (Figure 2). The details regarding the analysis of a high Ki-67 PI are shown in Table 3. Representative examples of low and high Ki-67 PI cases, along with their corresponding R2* maps, are shown in Figure 3.

Figure 2 ROC curve for the combination of hepatic R2* value (≥136 s−1) and AFP level (>400 ng/mL) in predicting a high Ki-67 proliferation index (>20%) in HCC. AFP, alpha-fetoprotein; AUC, area under the ROC curve; CI, confidence interval; HCC, hepatocellular carcinoma; ROC, receiver operating characteristic.

Figure 3 Representative cases of HCC lesions (indicated by arrowheads) with low (≤20%) or high (>20%) Ki-67 PI, along with their corresponding R2* maps. (A-C) A 65-year-old male with a 50-mm tumor in liver segment VI. (D-F) A 65-year-old male with a 20-mm tumor in liver segments VI and VII. (G-I) A 66-year-old female with a 27-mm tumor in the left hepatic lobe. AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma; PI, proliferation index; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging.

Predictors of CK19-positive expression

Univariate analysis identified the significant predictors of positive CK19 expression to be AFP >400 ng/mL (OR =8.42; 95% CI: 2.60–29.25; P<0.001), the difference between hepatic and intratumoral R2* values (OR =0.99; 95% CI: 0.98–1.00; P=0.061), and tumor peripheral washout (OR =6.48, 95% CI: 1.15–42.39; P=0.035). In multivariate analysis, the only independent predictor for CK19 positivity was AFP >400 ng/mL (OR =8.50; 95% CI: 2.38–30.34, P=0.001), while R2* value was not found to be a significant predictor (P=0.124). The details of the analysis on CK19 expression are provided in Table 4.

Interreader agreement

In the interreader agreement analysis, the ICC for intratumoral R2* measurements and hepatic R2* measurements was 0.954 (95% CI: 0.930–0.971; P<0.01) and 0.991 (95% CI: 0.986–0.994; P<0.01), respectively, indicating excellent consistency.

Discussion

Accurate preoperative prediction of HCC’s biological behavior is crucial for individualized treatment decision-making. In our retrospective single-center study, we used MRI R2* imaging to clarify the association between hepatic and intratumoral R2* values and the expression of Ki-67 and CK19 in patients with iron-sparing HCC. It was found that the hepatic R2* value was significantly lower in patients with a high Ki-67 PI (P=0.038). The combination of hepatic R2* value ≥136 s⁻¹ and AFP >400 ng/mL demonstrated good predictive ability for high Ki-67 PI, with an AUC of 0.771 (95% CI: 0.664–0.858). AFP >400 ng/mL was identified as an independent predictor for positive CK19 expression, while a hepatic R2* value ≥136 s⁻¹ demonstrated limited predictive ability (P=0.124).

The liver plays a central role in systemic iron metabolism, and hepatic disease may disrupt intracellular iron homeostasis, thereby altering iron metabolism in the liver. Previous research suggests that iron metabolism abnormalities vary across different stages of liver disease (36). In our study, hepatic iron content was an independent predictor of a high Ki-67 PI. However, the high Ki-67 PI group exhibited lower hepatic R2* values compared to the low Ki-67 PI group (P=0.038). One possible explanation for this is that highly proliferative tumors have higher iron requirements, which may be accompanied by the relative depletion of systemic iron stores. Since the liver is the primary iron storage organ, its iron content is closely linked to total whole-iron status (37,38), and R2* values exhibit a strong linear correlation with hepatic iron concentration (33,39). Consequently, lower hepatic R2* values in tumors with higher proliferative activity may reflect altered iron metabolism in the host-tumor setting. These findings are consistent with the study by Huang et al. (30). Other work (36,40) has shown that the serum iron levels of patients with HBV-related HCC are lower compared to those with HBV-related chronic hepatitis B (CHB) or HBV-related liver cirrhosis. Moreover, Wei et al. (40) found that serum iron levels are negatively correlated with tumor size. These findings constitute indirect support for an association between liver iron depletion and more aggressive tumor biology in HCC.

Although hepatic R2* ≥136 s⁻¹ remained a significant predictor for a high Ki-67 PI, its effect size was modest, and the discrimination ability of the combined model was only moderate. Therefore, this finding should be interpreted as exploratory and hypothesis-generating rather than clinically actionable. Serological markers such as ferritin and transferrin represent the simplest means of assessing systemic iron status (41), yet data related to these markers were unavailable for most participants in this study. Consequently, the hypothesis that tumors with a high Ki-67 PI deplete systemic iron and reduce hepatic R2* values has not yet been adequately tested, as no further validation was performed. Given these limitations, further validation in larger independent cohorts will be required to determine whether this association reflects a reproducible biological signal or a model-dependent finding.

Hepatic R2* values ≥136 s⁻¹ showed limited ability for predicting positive CK19 expression. The discrepancy in performance of hepatic R2* for Ki-67 and CK19 may reflect the biological differences between these markers. Ki-67 is a marker of proliferative activity (5) and may be more closely related to tumor iron demand and host iron utilization, whereas CK19 reflects a progenitor/biliary phenotype rather than proliferative burden per se (9). It is therefore plausible that hepatic iron-related measures show a modest association with Ki-67 but a limited association with CK19. In addition, the relatively small number of CK19-positive tumors in our cohort reduced the statistical power of the analysis.

LI‑RADS v2018 lists intratumoral ironsparing as an ancillary malignant feature (28). Ironfree nodules have been demonstrated to possess precancerous properties in cases of hemochromatosis (31,42). In our study, no significant differences in intratumoral R2* values were observed between the high and low Ki-67 PI groups (P=0.685) or between the CK19-positive and CK19-negative groups (P>0.99), consistent with the findings of Huang et al. (30). Additionally, no statistically significant difference was observed in intratumoral R2* values across the differentiation groups (P=0.248). Similarly, there were no significant differences in R2* values between HCC lesions with and without hemorrhage (P=0.787) or between those with and without necrosis (P=0.363). This may be attributed to malignant hepatocytes losing their iron storage capacity, resulting in “iron resistance” (43,44). In our cohort, the median tumor R2* value was 34.95 (IQR 21.67–49.93) s⁻¹, corresponding to a median iron concentration of approximately 0.44 mg/g (IQR, 0.26–0.64 mg/g) (33). These findings are consistent with those reported by Huang et al. (median R2* value 38.55 s⁻¹, IQR, 26.40–47.81 s⁻¹) (30).

Serum AFP is a commonly used biomarker for HCC, employed in the surveillance of high-risk populations and the assessment of recurrence (2). An AFP level >400 ng/mL is associated with rapid tumor growth (35) and serves as a predictor of posttreatment recurrence and prognosis (45,46). In our study, AFP >400 ng/mL was an independent predictor of a high Ki-67 PI and positive CK19 expression, a finding consistent with previous research (47-49).

From a clinical perspective, these results suggest that hepatic R2* mapping may complement serum AFP in the preoperative assessment of proliferative activity in patients with solitary iron-sparing HCC. R2* maps can be generated from routine multi-echo gradient-echo MRI without invasive sampling, and AFP has already been incorporated into standard protocols of HCC surveillance and prognostic assessment. Therefore, the combined use of hepatic R2* and AFP may help identify patients with a higher probability of an elevated Ki-67 PI before surgery, thus supporting more focused multidisciplinary discussion of surgical strategy, recurrence surveillance intensity, and postoperative risk assessment. In contrast, the absence of an independent association between R2* parameters and positive CK19 expression indicates that R2* mapping should not be used as a stand-alone surrogate for CK19 status.

This study involved several limitations that should be addressed. First, as we employed a retrospective single-center design and a relatively small sample size, the findings should be considered exploratory and remain to be validated in larger, prospective, multicenter, and multi-ethnic cohorts before broader conclusions can be drawn. Second, the generalizability of our findings may be limited by the characteristics of the study population. Our cohort primarily consisted of HBV-infected patients (86.25%) and male participants (93.75%). Therefore, whether these results apply to patients with HCV- or nonalcoholic steatohepatitis/nonalcoholic fatty liver disease-related HCC remains unclear. Third, intratumoral iron-sparing was used as an imaging-based selection criterion to define a specific HCC subgroup rather than as an independent biomarker applicable to all patients with HCC. We acknowledge that defining iron-sparing according to the relative tumor-to-background R2* signal introduces design dependence and limits the generalization of the findings beyond the selected iron-sparing phenotype. In addition, no universally accepted absolute R2* threshold is currently available for defining intratumoral iron-sparing in HCC. Fourth, intratumoral R2* was measured on the largest tumor section to maximize lesion conspicuity and ROI size, in accordance with previously reported methods (50,51). However, this single-slice approach may not fully capture intratumoral heterogeneity. Given that the source R2* maps had a slice thickness of 6–8 mm, finer interval sampling across adjacent levels was not feasible in the dataset. Fifth, although Ki-67 and CK19 expression are associated with prognosis, prognostic analysis was not performed due to insufficient follow-up duration in most of the included cases.

Conclusions

This retrospective single-center study of patients with solitary iron-sparing HCC produced data indicating that hepatic R2* (≥136 s⁻¹) combined with AFP (>400 ng/mL) may serve as a biomarker for predicting HCC with a high Ki-67 PI. This finding could help further clarify the relationship between liver iron content and proliferative tumor biology, providing a reference for optimizing therapeutic stratification. This study was the first to investigate the association between hepatic and intratumoral R2* levels with CK19 expression in patients with HCC. Although no statistically significant correlation was found, our findings may provide insights for future research.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0434/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0434/dss

Funding: This study was supported by grants from the National Natural Science Foundation of China (Nos. 82471567 and 81801757), and Guangdong Basic and Applied Basic Research Foundation (Nos. 2024A1515010279 and 2023A1515010256).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0434/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This retrospective single-center study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of The Third Affiliated Hospital of Sun Yat-sen University (approval No. II2025-229-01), and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Marrero JA, Kulik LM, Sirlin CB, Zhu AX, Finn RS, Abecassis MM, Roberts LR, Heimbach JK. Diagnosis, Staging, and Management of Hepatocellular Carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases. Hepatology 2018;68:723-50. [Crossref] [PubMed]
3. Portolani N, Coniglio A, Ghidoni S, Giovanelli M, Benetti A, Tiberio GA, Giulini SM. Early and late recurrence after liver resection for hepatocellular carcinoma: prognostic and therapeutic implications. Ann Surg 2006;243:229-35. [Crossref] [PubMed]
4. Chen J, Zhou J, Kuang S, Zhang Y, Xie S, He B, Deng Y, Yang H, Shan Q, Wu J, Sirlin CB, Wang J. Liver Imaging Reporting and Data System Category 5: MRI Predictors of Microvascular Invasion and Recurrence After Hepatectomy for Hepatocellular Carcinoma. AJR Am J Roentgenol 2019;213:821-30. [Crossref] [PubMed]
5. Shi W, Hu J, Zhu S, Shen X, Zhang X, Yang C, Gao H, Zhang H. Expression of MTA2 and Ki-67 in hepatocellular carcinoma and their correlation with prognosis. Int J Clin Exp Pathol 2015;8:13083-9.
6. Zhang L, Xiao Y, Dong M, Li M, Chen H, Wang J. Three-dimensional MR elastography-based stiffness for assessing the status of Ki67 proliferation index and Cytokeratin-19 in hepatocellular carcinoma. Eur Radiol 2025;35:4722-35. [Crossref] [PubMed]
7. Li HH, Qi LN, Ma L, Chen ZS, Xiang BD, Li LQ. Effect of KI-67 positive cellular index on prognosis after hepatectomy in Barcelona Clinic Liver Cancer stage A and B hepatocellular carcinoma with microvascular invasion. Onco Targets Ther 2018;11:4747-54. [Crossref] [PubMed]
8. Luo Y, Ren F, Liu Y, Shi Z, Tan Z, Xiong H, Dang Y, Chen G. Clinicopathological and prognostic significance of high Ki-67 labeling index in hepatocellular carcinoma patients: a meta-analysis. Int J Clin Exp Med 2015;8:10235-47.
9. Durnez A, Verslype C, Nevens F, Fevery J, Aerts R, Pirenne J, Lesaffre E, Libbrecht L, Desmet V, Roskams T. The clinicopathological and prognostic relevance of cytokeratin 7 and 19 expression in hepatocellular carcinoma. A possible progenitor cell origin. Histopathology 2006;49:138-51.
10. Han S, Meng F, Zhang HK, Li HL, Qu JR. Correlation analysis of Ki67, Ck19 with clinicopathological features and apparent diffusion coefficient value of hepatocellular carcinoma. Zhonghua Yi Xue Za Zhi 2021;101:798-802. [Crossref] [PubMed]
11. Kim H, Choi GH, Na DC, Ahn EY, Kim GI, Lee JE, Cho JY, Yoo JE, Choi JS, Park YN. Human hepatocellular carcinomas with "Stemness"-related marker expression: keratin 19 expression and a poor prognosis. Hepatology 2011;54:1707-17. [Crossref] [PubMed]
12. Govaere O, Komuta M, Berkers J, Spee B, Janssen C, de Luca F, et al. Keratin 19: a key role player in the invasion of human hepatocellular carcinomas. Gut 2014;63:674-85. [Crossref] [PubMed]
13. Lu XY, Xi T, Lau WY, Dong H, Zhu Z, Shen F, Wu MC, Cong WM. Hepatocellular carcinoma expressing cholangiocyte phenotype is a novel subtype with highly aggressive behavior. Ann Surg Oncol 2011;18:2210-7. [Crossref] [PubMed]
14. Choi SY, Kim SH, Park CK, Min JH, Lee JE, Choi YH, Lee BR. Imaging Features of Gadoxetic Acid-enhanced and Diffusion-weighted MR Imaging for Identifying Cytokeratin 19-positive Hepatocellular Carcinoma: A Retrospective Observational Study. Radiology 2018;286:897-908. [Crossref] [PubMed]
15. Takahashi Y, Dungubat E, Kusano H, Ganbat D, Tomita Y, Odgerel S, Fukusato T. Application of Immunohistochemistry in the Pathological Diagnosis of Liver Tumors. Int J Mol Sci 2021;22:5780. [Crossref] [PubMed]
16. Robert M, Sofair AN, Thomas A, Bell B, Bialek S, Corless C, Van Ness G, Huie-White S, Stabach N, Zaman A. A comparison of hepatopathologists' and community pathologists' review of liver biopsy specimens from patients with hepatitis C. Clin Gastroenterol Hepatol 2009;7:335-8. [Crossref] [PubMed]
17. Liu Z, Yang S, Chen X, Luo C, Feng J, Chen H, Ouyang F, Zhang R, Li X, Liu W, Guo B, Hu Q. Nomogram development and validation to predict Ki-67 expression of hepatocellular carcinoma derived from Gd-EOB-DTPA-enhanced MRI combined with T1 mapping. Front Oncol 2022;12:954445. [Crossref] [PubMed]
18. Qin Q, Deng LP, Chen J, Ye Z, Wu YY, Yuan Y, Song B. The value of MRI in predicting hepatocellular carcinoma with cytokeratin 19 expression: a systematic review and meta-analysis. Clin Radiol 2023;78:e975-84. [Crossref] [PubMed]
19. Huang J, Shen D, Sun J, Shi J, Dou W, Prince M, Ye J, Luo X. Amide proton transfer-weighted imaging in predicting aggressiveness of hepatocellular carcinoma: comparison with diffusion-weighted imaging. Quant Imaging Med Surg 2026;16:88. [Crossref] [PubMed]
20. Yang NJ, Zeng WT, Xiang HY, Wáng YXJ. Classification of hepatocellular carcinoma Ki-67 expression status by a simple combination of magnetic resonance slow diffusion coefficient (SDC) and apparent diffusion coefficient (ADC): initial promising results. Quant Imaging Med Surg 2026;16:500. [Crossref] [PubMed]
21. Wang F, Yan C, Huang X, He J, Yang M, Xian D. Radiomics and Deep Learning as Important Techniques of Artificial Intelligence - Diagnosing Perspectives in Cytokeratin 19 Positive Hepatocellular Carcinoma. J Hepatocell Carcinoma 2025;12:1129-40. [Crossref] [PubMed]
22. Zhou L, Chen Y, Li Y, Wu C, Xue C, Wang X. Diagnostic value of radiomics in predicting Ki-67 and cytokeratin 19 expression in hepatocellular carcinoma: a systematic review and meta-analysis. Front Oncol 2023;13:1323534. [Crossref] [PubMed]
23. Zuo XY, Liu HF. Biparametric magnetic resonance imaging-based radiomic and deep learning models for predicting Ki-67 risk stratification in hepatocellular carcinoma. World J Hepatol 2025;17:109530. [Crossref] [PubMed]
24. Cassat JE, Skaar EP. Iron in infection and immunity. Cell Host Microbe 2013;13:509-19. [Crossref] [PubMed]
25. Ba Q, Hao M, Huang H, Hou J, Ge S, Zhang Z, Yin J, Chu R, Jiang H, Wang F, Chen K, Liu H, Wang H. Iron deprivation suppresses hepatocellular carcinoma growth in experimental studies. Clin Cancer Res 2011;17:7625-33. [Crossref] [PubMed]
26. Niederau C, Fischer R, Pürschel A, Stremmel W, Häussinger D, Strohmeyer G. Long-term survival in patients with hereditary hemochromatosis. Gastroenterology 1996;110:1107-19. [Crossref] [PubMed]
27. Urano S, Ohara T, Noma K, Katsube R, Ninomiya T, Tomono Y, Tazawa H, Kagawa S, Shirakawa Y, Kimura F, Nouso K, Matsukawa A, Yamamoto K, Fujiwara T. Iron depletion enhances the effect of sorafenib in hepatocarcinoma. Cancer Biol Ther 2016;17:648-56. [Crossref] [PubMed]
28. Cerny M, Chernyak V, Olivié D, Billiard JS, Murphy-Lavallée J, Kielar AZ, Elsayes KM, Bourque L, Hooker JC, Sirlin CB, Tang A. LI-RADS Version 2018 Ancillary Features at MRI. Radiographics 2018;38:1973-2001. [Crossref] [PubMed]
29. Reeder SB, Yokoo T, França M, Hernando D, Alberich-Bayarri Á, Alústiza JM, Gandon Y, Henninger B, Hillenbrand C, Jhaveri K, Karçaaltıncaba M, Kühn JP, Mojtahed A, Serai SD, Ward R, Wood JC, Yamamura J, Martí-Bonmatí L. Quantification of Liver Iron Overload with MRI: Review and Guidelines from the ESGAR and SAR. Radiology 2023;307:e221856. [Crossref] [PubMed]
30. Huang M, Zhang F, Li Z, Luo Y, Li J, Wang Z, Ma L, Chen G, Hu X. Fat fraction quantification with MRI estimates tumor proliferation of hepatocellular carcinoma. Front Oncol 2024;14:1367907. [Crossref] [PubMed]
31. Deugnier YM, Charalambous P, Le Quilleuc D, Turlin B, Searle J, Brissot P, Powell LW, Halliday JW. Preneoplastic significance of hepatic iron-free foci in genetic hemochromatosis: a study of 185 patients. Hepatology 1993;18:1363-9.
32. Hong CW, Hamilton G, Hooker C, Park CC, Tran CA, Henderson WC, Hooker JC, Fazeli Dehkordy S, Schwimmer JB, Reeder SB, Sirlin CB. Measurement of spleen fat on MRI-proton density fat fraction arises from reconstruction of noise. Abdom Radiol (NY) 2019;44:3295-303. [Crossref] [PubMed]
33. Hernando D, Zhao R, Yuan Q, Aliyari Ghasabeh M, Ruschke S, Miao X, Karampinos DC, Mao L, Harris DT, Mattison RJ, Jeng MR, Pedrosa I, Kamel IR, Vasanawala S, Yokoo T, Reeder SB. Multicenter Reproducibility of Liver Iron Quantification with 1.5-T and 3.0-T MRI. Radiology 2023;306:e213256. [Crossref] [PubMed]
34. Chernyak V, Fowler KJ, Kamaya A, Kielar AZ, Elsayes KM, Bashir MR, Kono Y, Do RK, Mitchell DG, Singal AG, Tang A, Sirlin CB. Liver Imaging Reporting and Data System (LI-RADS) Version 2018: Imaging of Hepatocellular Carcinoma in At-Risk Patients. Radiology 2018;289:816-30. [Crossref] [PubMed]
35. Jang HJ, Choi SH, Wee S, Choi SJ, Byun JH, Won HJ, Shin YM, Sirlin CB. CT- and MRI-based Factors Associated with Rapid Growth in Early-Stage Hepatocellular Carcinoma. Radiology 2024;313:e240961. [Crossref] [PubMed]
36. Gao YH, Wang JY, Liu PY, Sun J, Wang XM, Wu RH, He XT, Tu ZK, Wang CG, Xu HQ, Niu JQ. Iron metabolism disorders in patients with hepatitis B-related liver diseases. World J Clin Cases 2018;6:600-10. [Crossref] [PubMed]
37. Angelucci E, Brittenham GM, McLaren CE, Ripalti M, Baronciani D, Giardini C, Galimberti M, Polchi P, Lucarelli G. Hepatic iron concentration and total body iron stores in thalassemia major. N Engl J Med 2000;343:327-31. [Crossref] [PubMed]
38. European Association For The Study Of The Liver. EASL clinical practice guidelines for HFE hemochromatosis. J Hepatol 2010;53:3-22. [Crossref] [PubMed]
39. Hankins JS, McCarville MB, Loeffler RB, Smeltzer MP, Onciu M, Hoffer FA, Li CS, Wang WC, Ware RE, Hillenbrand CM. R2* magnetic resonance imaging of the liver in patients with iron overload. Blood 2009;113:4853-5. [Crossref] [PubMed]
40. Wei Y, Ye W, Zhao W. Serum Iron Levels Decreased in Patients with HBV-Related Hepatocellular Carcinoma, as a Risk Factor for the Prognosis of HBV-Related HCC. Front Physiol 2018;9:66. [Crossref] [PubMed]
41. Wang W, Knovich MA, Coffman LG, Torti FM, Torti SV. Serum ferritin: Past, present and future. Biochim Biophys Acta 2010;1800:760-9. [Crossref] [PubMed]
42. Guyader D, Gandon Y, Sapey T, Turlin B, Mendler MH, Brissot P, Deugnier Y. Magnetic resonance iron-free nodules in genetic hemochromatosis. Am J Gastroenterol 1999;94:1083-6. [Crossref] [PubMed]
43. Terada T, Nakanuma Y. Iron-negative foci in siderotic macroregenerative nodules in human cirrhotic liver. A marker of incipient neoplastic lesions. Arch Pathol Lab Med 1989;113:916-20.
44. Zhang J, Krinsky GA. Iron-containing nodules of cirrhosis. NMR Biomed 2004;17:459-64. [Crossref] [PubMed]
45. Kim DH, Choi SH, Koo B, Choi SJ, Jang HJ, Heo S, Byun JH, Won HJ, Shin YM. Effect of combining serum alpha-fetoprotein with LI-RADS v2018 on gadoxetate-enhanced MRI in the diagnosis and prognostication of hepatocellular carcinoma. Eur Radiol 2025;35:4957-66. [Crossref] [PubMed]
46. Wei H, Jiang H, Liu X, Qin Y, Zheng T, Liu S, Zhang X, Song B. Can LI-RADS imaging features at gadoxetic acid-enhanced MRI predict aggressive features on pathology of single hepatocellular carcinoma? Eur J Radiol 2020;132:109312. [Crossref] [PubMed]
47. Zhao Y, Tan X, Chen J, Tan H, Huang H, Luo P, Liang Y, Jiang X. Preoperative prediction of cytokeratin-19 expression for hepatocellular carcinoma using T1 mapping on gadoxetic acid-enhanced MRI combined with diffusion-weighted imaging and clinical indicators. Front Oncol 2022;12:1068231. [Crossref] [PubMed]
48. Qiu G, Chen J, Liao W, Liu Y, Wen Z, Zhao Y. Gadoxetic acid-enhanced MRI combined with T1 mapping and clinical factors to predict Ki-67 expression of hepatocellular carcinoma. Front Oncol 2023;13:1134646. [Crossref] [PubMed]
49. Gu Y, Jin K, Gao S, Sun W, Yin M, Han J, Zhang Y, Wang X, Zeng M, Sheng R. A preoperative nomogram with MR elastography in identifying cytokeratin 19 status of hepatocellular carcinoma. Br J Radiol 2025;98:210-9. [Crossref] [PubMed]
50. Kitagawa T, Kozaka K, Matsubara T, Wakayama T, Takamatsu A, Kobayashi T, Okumura K, Yoshida K, Yoneda N, Kitao A, Kobayashi S, Gabata T, Matsui O, Heiken JP. Fat fraction and R2 * values of various liver masses: Initial experience with 6-point Dixon method on a 3T MRI system. Eur J Radiol Open 2023;11:100519. [Crossref] [PubMed]
51. Kupczyk PA, Kurt D, Endler C, Luetkens JA, Kukuk GM, Fronhoffs F, Fischer HP, Attenberger UI, Pieper CC. MRI proton density fat fraction for estimation of tumor grade in steatotic hepatocellular carcinoma. Eur Radiol 2023;33:8974-85. [Crossref] [PubMed]