MRI Features and Neutrophil-to-Lymphocyte Ratio (NLR)-Based Nomogram to Predict Prognosis of Microvascular Invasion-Negative Hepatocellular Carcinoma

Introduction

Liver cancer is the sixth most common malignancy and the third leading cause of cancer-related death worldwide, according to 2022 global cancer statistics.¹ Hepatocellular carcinoma (HCC) accounts for 75–85% of liver cancer. For HCC patients, hepatic resection has been the mainstay of curative treatment. But the 5-year recurrence rate after surgery is as high as 70%, which is the main cause of poor prognosis for HCC patients.²

Microvascular invasion (MVI) is a robust predictor for HCC recurrence after hepatectomy.³ It plays a crucial role in therapeutic decision-making, such as opting for surgery instead of ablation,⁴ selecting a wider surgical margin at hepatectomy, more intensive postoperative monitoring,⁵ and considering adjuvant therapies.⁶ However, MVI has been reported in only about 15–57% of HCC,^4–7 and many MVI-negative patients still experience postoperative recurrence in clinical practice. This suggests that research on the prognosis of MVI-negative HCC patients should be carried out to help the clinic fully understand and deal with the potential recurrence risk.

MRI has been widely used for the diagnosis and assessment of HCC. Prior works explored some imaging features, such as tumor size, tumor capsule, mosaic architecture, and Liver Imaging Reporting and Data System (LI-RADS), which can serve as noninvasive risk factors to evaluate the aggressive biological behavior of HCC.^7–9 Additionally, systemic inflammation is a key risk factor for HCC progression and poor outcomes. Inflammation can enhance tumor progression and immune evasion.^10,11 Therefore, an HCC-promoting state can be suspected from high levels of inflammatory markers, thus resulting in a poor prognosis. Inflammatory markers like the neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), lymphocyte-to-monocyte ratio (LMR), and systemic immune inflammation index (SII), have been correlated with worse survival outcomes in HCC patients.^12–14 But to the best of our knowledge, only a few studies have explored the relationship between MRI imaging features and prognosis in MVI-negative HCC patients.^15–18 Moreover, these studies have been limited by their short follow-up periods and a lack of emphasis on inflammatory markers. It’s also important to note that the nomogram, a tool that combines various prognostic and decisive factors to calculate individual numerical probabilities for clinical outcomes, satisfies the clinical need for comprehensive biological and clinical models.¹⁹ It also aligns with the goal of personalized medicine and has been extensively used in the prognostic evaluation of HCC patients.^15–18

Therefore, we hypothesized that the certain MRI features and inflammatory markers were associated with recurrence-free survival (RFS) of MVI-negative HCC patients. We believed that when these factors were incorporated into a nomogram model, it could accurately predict prognosis and offer significant clinical application.

Materials and Methods

Patients

This retrospective study performed at a single medical institution. Ethics committee approval was granted by our institutional review board, which waived the requirement for informed consent. And this study was conducted in accordance with the Declaration of Helsinki.

Between January 2014 and November 2018, a total of 143 MVI-negative HCC patients who underwent dynamic contrast-enhanced MRI followed by curative hepatectomy were included. The inclusion criteria were as follows: (1) patients with MVI-negative HCC confirmed by histopathologic examination; (2) patients treated by resection with tumor-negative resection margins (R0 resection); (3) MRI quality was adequate for analysis. The exclusion criteria were as follows: (1) patients have two or more HCC lesions; (2) HCC tumor size larger than 10 cm; (3) patients treated by prior cancer-related treatment before surgery; (4) patients with other malignant tumors; (5) those that were lost to death, and lost to follow-up within 5 years. Figure 1 displays the flowchart detailing the selection of eligible patients.

Figure 1 Flowchart of this study population. Abbreviations: HCC, hepatocellular carcinoma; MVI, microvascular invasion.

MRI Acquisition and Imaging Analysis

MRI was performed on a 1.5-T system (Magnetom Avanto, Siemens Healthcare, Erlangen, Germany). Baseline MRI included a T1-weighted sequence, a T2-weighted and T2-weighted fat-suppressed sequence. For gadolinium- diethylenetriamine pentaacetic acid (Gd-DTPA, 0.1 mmol/kg of body weight; Magnevist, Schering, Berlin, Germany)-enhanced MRI, arterial phase (25−30 seconds), portal venous phase (55−65 seconds), and delayed phase (2 mins) were obtained by using a T1-weighted three-dimensional turbo-field-echo sequence. The acquisition parameters are listed in Supplementary Table S1.

Images were independently analyzed by two radiologists (H.L. and Y.W, with 16 and 6 years of experience in abdominal imaging, respectively) who were blinded to clinical, histological, and laboratory, and follow-up results. For discordant cases, the two reviewers met to discuss and draw final conclusions by consensus.

The two reviewers independently evaluated some imaging features for each HCC as defined in LI-RADS v2018,²⁰ including major imaging features (including tumor size, non-rim arterial phase hyperenhancement [APHE], non-peripheral washout, and enhancing capsule), ancillary imaging features (including mosaic architecture, nodule-in-nodule architecture, fat in mass, and blood products in mass), two non-LI-RADS imaging features (non-smooth tumor margin and tumor capsule).

Nonsmooth tumor margin was defined as non-nodular tumors with extranodular extension that had a budding portion at its periphery protruding into the liver parenchyma or multinodular confluence appearance.⁸ Complete tumor capsule was defined as complete peripheral rim of smooth hyperenhancement in the portal venous or delayed phase. Incomplete or without tumor capsule was defined as the absence of “capsule” or the presence of a disrupted “capsule” in any imaging plane.²¹ Enhancing capsule was defined as a linear enhancing rim surrounding the tumor in the portal venous or delayed phase, including both complete and incomplete rims.²²

Clinical Data Acquisition

The following clinical data were analyzed: age; gender; drink; HBsAg, body mass index (BMI); liver cirrhosis and Edmondson-Steiner grade. In addition, the laboratory results were collected, including: alpha-fetoprotein (AFP; < 400 or ≥ 400 ng/mL); neutrophil-to-lymphocyte ratio (NLR; < 2 or ≥ 2); platelet-to-lymphocyte ratio (PLR; < 150 or ≥ 150); lymphocyte-to-monocyte ratio (LMR; < 4 or ≥ 4); systemic immune inflammation index (SII); white blood cell (WBC); platelet (PLT; < 100 or ≥ 100); neutrophils (Neu); alanine aminotransferase (ALT; ≤ 40 or > 40 U/L); and aspartate aminotransferase (AST; ≤ 40 or > 40 U/L).

Follow-Up

Postoperatively, all patients received standardized clinical and radiologic follow-up according to the institutional protocol. The surveillance program consisted of laboratory tests including AFP, liver function and ultrasound examination, performed every 1 month for the first year after surgery and then every 3 months. Additionally, contrast-enhanced CT or MRI was performed every 3–6 months. The primary end point was tumor recurrence. RFS was calculated from the date of curative resection to the date of recurrence or death or the most recent follow-up. The mean follow-up period was 55 months (range, 2–96 months).

Statistical Analysis

The Chi-square or Fisher’s exact test were used to compare categorical variables. The Mann–Whitney U-test was used for continuous variables. Independent prognostic factors of RFS were identified by univariate and multivariate Cox proportional hazards regression analysis. Subsequently, a nomogram was constructed on the results of multivariable Cox regression analysis. The receiver operating characteristic (ROC) curve, concordance index (C-index) and calibration curve were used to evaluate the predictive accuracy and discriminative ability of the nomogram. Each patient was assigned a total risk score (NomoScore: nomogram risk score) for risk stratification of RFS according to the nomogram model. Patients were divided into different risk groups (low- or high-risk group) to identify significant differences in RFS between the groups. Decision curve analysis (DCA) was conducted to determine the clinical benefit of the nomogram by quantifying the net benefit along with the increase in threshold probabilities.

All statistical analyses were performed by using statistical SPSS (version 22.6; IBM Corp., Armonk, NY, USA); and R version 3.6.3 (http://www.r-project.org/). Differences with a P-value < 0.05 were considered statistically significant.

Results

Clinical Characteristics and Recurrence

Baseline and clinicopathological features of 143 patients with solitary MVI-Negative HCC Patients (113 males and 30 females; mean age: 51.4 ± 11.3 years; range: 24–76 years) are summarized in Table 1.

Table 1 Baseline and Clinicopathological Characteristics of MVI-Negative HCC Patients

At the end of the follow-up period, 50 patients experienced recurrence and 93 patients were free of recurrence. The median follow-up time was 55 months (2–96 months). The cumulative 1-, 3-, and 5-year RFS rates were 86%, 75%, and 68%, respectively.

After comparing the baseline data of recurrence group (n = 50) and recurrence-free group (n = 93), correlation analysis showed that there were significant differences in drink (P = 0.028), NLR ≥ 2 (P = 0.014), tumor size (P = 0.010), tumor capsule (P = 0.001), and mosaic architecture (P = 0.004) between the groups (Table 1).

Development and Validation of Nomogram Prediction Model

In univariable Cox analysis, postoperative recurrence was associated with five predictors: NLR ≥ 2 (P = 0.014), drink (P = 0.049), tumor size (P = 0.001), tumor capsule (P = 0.000), and mosaic architecture (P = 0.001) (Table 2). In multivariable analysis, NLR ≥ 2 (hazard ratio [HR]: 1.999; P = 0.018), tumor size (HR: 1.236; P = 0.002), and tumor capsule (HR: 3.409; P = 0.000) were independent significant variables associated with tumor recurrence (Table 2) (Figures 2 and 3).

Table 2 Univariate and Multivariate Cox Analyses for RFS of MVI-Negative HCC Patients

Figure 2 MRI images of MVI-negative HCC patients who did not relapse during follow-up. (A) Delayed phase image of a 44-year-old male patient showing a 3.5 cm mass (arrow) with complete capsule. Tumor recurrence has not been found after 70 months of follow-up. (B) Delayed phase image of a 60-year-old male patient showing a 2.6 cm mass (arrow) without capsule. Tumor recurrence has not been found after 71 months of follow-up.

Figure 3 MRI images of MVI-negative HCC patients who relapsed within 5 years after surgery. (A) Delayed phase image of a 68-year-old male patient showing a 4.7 cm mass (arrow) with incomplete capsule. Tumor recurrence was detected 15 months after curative resection. (B) Delayed phase image of a 58-year-old male patient showing a 4.7 cm mass (arrow) with incomplete capsule. Tumor recurrence was detected 45 months after curative resection.

A Cox regression model-based nomogram was developed from these significant independent variables (Figure 4). The nomogram achieved a good C-index of 0.730 (95% CI: 0.656–0.804) by using bootstrapping to measure the discrimination in predicting 1-, 3-, and 5-year RFS rates. To evaluate the performance of the nomogram, the ROC curves were generated and the areas under curve (AUCs) were calculated. In ROC analysis, the AUCs of the nomogram for 1-, 3- and 5-year RFS prediction were 0.725, 0.784 and 0.798, respectively (Figure 5A). The calibration curves for 1-, 3- and 5-year RFS rates were largely overlapped with the standard lines (Figure 5B–D). The final nomogram could accurately distinguish the RFS of MVI-negative HCC patients and had better consistency between the predicted probability and the observed probability of RFS.

Figure 4 Nomogram for predicting 1-, 3-, and 5-year recurrence-free survival (RFS) in MVI-negative HCC patients after hepatectomy. Each independent prognostic factor identified in the Cox regression was assigned a point (complete capsule is assigned 0 point and incomplete/without capsule is assigned 58 points, neutrophil-to-lymphocyte ratio [NLR] < 2 is assigned 0 point and NLR ≥ 2 is assigned 33 points, tumor size is assigned 10 points per 1 cm). The total points could be obtained by calculating the sum of all factors. With the total points, the probabilities of 1-, 3- and 5-year RFS could be predicted.

Figure 5 Evaluation of the performance of the nomogram. (A) Receiver operating characteristic curve for the nomogram model. (B–D) Calibration curves of nomogram for predicting 1-, 3-, and 5-year probability of recurrence-free survival (RFS).

Prognostic Assessment and Risk Stratification

MVI-negative HCC patients were divided into high-risk group (NomoScore ≥ 109) or the low-risk group (NomoScore < 109) based on the total risk scores calculated by the nomogram model. Survival curves were calculated to compare the RFS rates between the two groups using the Kaplan–Meier method, and the results showed a significant discriminatory ability. (P < 0.0001, Figure 6A). DCA analysis showed that the nomogram could increase net benefit and exhibited a wider range of threshold probabilities than the independent risk factors (NLR, tumor size and tumor capsule) by the risk stratification (Figure 6B–D).

Figure 6 Prognostic assessment and risk stratification of the nomogram prediction model. (A) Risk stratification for recurrence-free survival (RFS). (B–D) Decision curve analyses of 1-, 3-, and 5-year RFS rates show an increase in net benefit.

Discussion

The lack of prognostic studies for MVI-negative HCC patients after hepatectomy poses a great challenge to clinical management. Therefore, our study specifically focused on MVI-negative patients, a subgroup that has rarely been investigated separately before, and aimed at constructing a prognostic nomogram designed for them.

Our study demonstrated that tumor size, tumor capsule and NLR were independent significant variables associated with RFS of MVI-negative HCC. And we developed an easy-to-use and effective nomogram model, which showed good predictive performance. Furthermore, the nomogram model was capable of stratifying patients into 2 prognostically distinct risk groups. This was clinically relevant because it may help to identify the patients at high risk of recurrence, for whom closer surveillance could be considered, and adjuvant therapies might provide survival benefit.

Previous studies have also confirmed that tumor size plays an important role in the prognosis of MVI-negative HCC patients.^15,17 Although there is no uniformity in the criteria for predicting recurrence based on tumor size, it is generally accepted that, as the tumor size increases, the risk of intrahepatic metastasis increases.⁵

The tumor capsule, as an imaging feature, refers to the hyperplastic fibrous connective tissues surrounding the tumor. It is actually a “pseudo-capsule” because it lacks epithelial tissues. But it may help to limit the growth of tumor.²³ Our study found that incomplete or without tumor capsule is an independent predictor of RFS, which is consistent with reports by Wei et al and Ng et al^21,23,24 A possible reason for this may be that, incomplete or without tumor capsule is associated with high BRAF and RAF1 expression in HCC, which can promote tumor invasion and metastasis.²⁵

Interestingly, previous studies found that tumor size and tumor capsule are also indicators for MVI,^26–28 a significant risk factor of recurrence in HCC patients. Therefore, these imaging features can be used to predict MVI expression in HCC and thus to evaluate the prognosis of patients. However, in our study, we found a direct association between these two imaging features and postoperative recurrence in MVI-negative HCC patients. Hence, we propose that imaging features may reflect the gene expression patterns of the tumor, not just MVI. This perspective is consistent with earlier findings^3,29,30 and introduces a novel and innovative viewpoint that we have derived from similar researches.^16–18

Findings from this study also contribute to elucidate the clinical value of inflammatory marker. Elevated NLR plays a crucial part in the progression of various malignancies.^12,31,32 Neutrophils are thought to drive tumor progression by enhancing tumor cell survival, invasiveness and metastatic ability, immunosuppression, extracellular matrix remodeling, and angiogenesis.³³ While lymphocytes control tumor growth by secreting cytokines and inducing cytotoxic cell death, and the reduction of lymphocyte levels will damage the immune function of the host.³⁴ Moreover, high NLR might be linked with greater disease burden and liver dysfunction. Our study showed that NLR ≥ 2 is an independent prognosticator of poor RFS in MVI-negative HCC patients. This indicates that a higher NLR is associated with poorer survival outcomes, a finding that aligns with previous research.^12,14

In this study, the dataset was not splitted into two groups randomly, a training group and a validation group. On the contrary, we only developed the internal validation of the nomogram with a representative sample size to avoid data waste and statistical inefficiency.

There are limitations in this study. First, this is a single-center study, and the selection bias is inevitable owing to the retrospective design. Second, as we only enrolled patients with a solitary HCC ≤ 10 cm, the results cannot be generalized to multiple tumors or tumors larger than 10 cm. Third, there are still some disputes about the optimal cut-off value of NLR, PLR and LMR, but our study chose a more recognized cut-off value to improve the accuracy of the conclusion. Fourth, imaging features of the hepatobiliary phase (HBP) were not incorporated in this study due to limitation of the contrast agent, and thus could not be compared with some of the previous studies of MVI-negative HCC patients.^16,18 Therefore, future prospective study with multi-center and larger samples are essential to verify our results.

Conclusion

In conclusion, we constructed a nomogram to predict RFS after curative resection in MVI-negative HCC patients by combining MRI features and inflammatory marker. This predictive model can significantly improve the detection rate of high-risk patients, and assist clinicians in treatment decision-making.