Clinical impact of the Model for End Liver Disease (MELD) score on the presence of microvascular invasion and on the postoperative outcome in patients undergoing liver transplantation

ABSTRACT Objective: to correlate clinical and epidemiological data with the pathological analysis of liver explants from patients undergoing liver transplantation for hetapocarcinoma in the UNICAMP HC and to verify whether the MELD and MELD-Na scores are reliable factors to predict a worse post-transplant prognosis. Methods: we studied liver transplants carried out between May 2010 and November 2017. After excluding 38 patients, we included 87, analyzing clinical and laboratory data for correlation with the outcome Microvascular Invasion (MVI). Subsequently, we computed the MELD and MELD-Na scores and performed a descriptive analysis of clinical and laboratory data and, finally, calculated ROC curves to assess the association between these laboratory parameters and mortality in these patients. Results: most patients were male (78.30%), with an average age of 58.53 years. Most liver diseases were caused by HCV (53.26%). We found no predictors for MVI among the laboratory parameters. The ROC curves for death identified the MELD score as the cutoff point with the highest combined sensitivity (90.91%) and specificity (37.50%), with a value of 10 points, whereas in the MELD-Na the cutoff point was 7 points, with a sensitivity of 90.91% and a specificity of 33.33%, both scores being significant. Conclusions: there were no reliable predictors of MVI between clinical, laboratory, and epidemiological variables. The MELD-Na score is more sensitive than the MELD one for predicting mortality in patients undergoing liver transplantation.

go through the stage of liver fibrosis before culminating in hepatocellular carcinoma, which does not exclude the possibility of direct evolution to HCC, though 8 .
NAFLD has an estimated global prevalence of 25% and has become the most common cause of HCC, being among the second or third indication for liver transplantation 9,10 .Approximately one quarter of adults with NAFLD have non-alcoholic steatohepatitis (NASH), which usually leads to progressive liver fibrosis, cirrhosis, and eventually evolves to hepatocellular carcinoma [11][12][13] .
Studies report a steady increase in the prevalence and incidence of NAFLD-related hepatocellular carcinoma.Soon there will likely be more cases of HCC related to NASH than to chronic hepatitis C.This will be possible because the treatment for the C virus is effective and can achieve viral eradication in almost all patients, while NASH is increasingly prevalent [14][15][16] .
The treatment of choice for advanced liver failure is human liver transplantation, and this therapeutic modality is limited because the demand for this organ is greater than the supply.For this reason, there is a list of potential liver transplant recipients, the order being defined by clinical criteria that assess the risk of mortality in the transplant waiting list, such as the Model for End-Stage Liver Disease (MELD), adopted in Brazil in 2006 [17][18][19] .Some histological features of HCC are directly related to the disease evolution, such as the occurrence of microvascular invasion (MVI).Several authors have shown that the presence of MVI was closely related with HCC recurrence, with worst prognosis and with short survival, thus limiting the therapeutic approach options (20).However, MVI is difficult to detect by imaging and laboratory methods, requiring histopathological evaluation of the liver explant, restricting the usefulness of this information in the preoperative period.Therefore, the attending physician may have limited information to consider available therapies [21][22][23] .
Although the occurrence of MVI is still a poorly understood phenomenon, several studies have been carried out to understand the relationship with other clinical and epidemiological data 20,22,24,25 .Computed tomography (CT) analysis, for example, did not predict MVI and, to date, there are no other methods proven to be able to predict the occurrence of MVI in general 26 .
Studies have shown that serum levels of some markers, such as alpha-fetoprotein (AFP) and Gamma GT, can predict the presence of MVI in patients with multiple liver nodules 22 .
Therefore, predicting MVI or determining related clinical factors can help in the operative management and therapeutic planning, which can positively impact disease's prognosis and patient survival.In this regard, the present study will be based on a retrospective analysis of patients undergoing liver transplantation, to determine whether the etiology of liver disease was able to predict the occurrence of microvascular invasion, in addition to discussing the clinical and epidemiological characteristics of all transplant recipients.

GOALS
The aim of this study was to correlate clinical and epidemiological data with the pathological analysis of liver explants from all patients who underwent liver

DISCUSSION AND CONCLUSION
The development of theoretical-clinical models capable of predicting unfavorable outcomes helps in planning and managing the patient's treatment.
Although there are numerous methods able at predicting various illnesses, there is often a high level of difficulty in creating these tools.Great scientific production and distinct approaches to the same subject by different researchers are therefore needed.The present study aimed at correlating the clinical and laboratory information of patients undergoing liver transplantation due to HCC to verify whether there is a relationship between these data and MVI or death.Recently, other authors performed calculations using data obtained by multislice tomography and concluded that it was not possible to predict MVI, as Lahan et al. 26 , who concluded that quantitative tomographic analysis does not predict microvascular invasion.
In this study, we observed that the epidemiological characteristics of patients agree with the national profile.In Brazil, about 70% of HCC cases are related to cirrhosis secondary to infection by viruses B or C 27 .In addition, there is a predominance of virus C over B nationally, with 54% and 16%, respectively 27 .
In the present study, however, although most HCC Barbosa Clinical impact of the Model for End Liver Disease (MELD) score on the presence of microvascular invasion and on the postoperative outcome in patients undergoing liver transplantation cases were associated with virus C, the alcoholic cause is represented as the second most prevalent etiology, reaching 17%, while virus B was at 6% (Table 1).It is worth noting that the sample used in this study is smaller when compared to the study by Carrilho et al. 27 , which defined the Brazilian profile for HCC.This sample difference could justify the divergence between the second etiology of HCC being the alcoholic cause and not the B virus in our series.
Currently, the lifestyle and eating habits of patients has drawn attention to metabolic changes.
Soon the metabolic causes of cirrhosis may be more prevalent than the viral causes 28 .In addition, our study confirmed that alcoholism is also a matter of concern for hepatocellular carcinoma.
The result on the HCC etiologies was directly related to the lifestyle and/or comorbidities of the studied patients.After analyzing the medical records, we found that 70% of these patients reported chronic use of alcoholic beverages at some time prior to the transplant (alcoholic liver disease was the second most prevalent cause in this study).In addition, 30% had a diagnosis of diabetes mellitus and 53% were hypertensive, becoming apparent that most patients had some type of metabolic comorbidity, as recorded in Table 1.In addition, when compared to national data, patients in this study showed a higher percentage of co-infection by viruses B and C; while the country has a percentage of 2%, our patients displayed 6%.
The occurrence of liver disease due to an unknown (cryptogenic) cause, 16%, was also above the Brazilian prevalence, 3%.A possible reason could be NASH under-diagnosed cases.It is noteworthy that the study that described the Brazilian epidemiology dates to 2010 27 .Therefore, we may be facing a changing trend in the Brazilian epidemiological profile, since the anti viral therapies are advancing, while patients' lifestyle is not following suit 35 .
Table 1 records the clinical characteristics of transplant recipients, 78% being men, with a mean age of 58 years (36-81).In addition, Table 4 shows that among all pre-clinical variables analyzed, male sex was a protective factor for the outcome death.Thus, although more men are affected by HCC (78%), a lower percentage died when compared with women.Females, however, were in a smaller number in our study, 21 (22%), which could justify this finding.On the other hand, the percentage of males (78%) was the same (78%) as in the study by Carrilho et al. 27 , who analyzed 1,363 patients with hepatocellular carcinoma in 2010.
This information reveals that there was no trend towards a reduction in HCC in men, the rate remaining constant.Similarly, as seen in Table 4, we did not find any relationship between the analyzed parameters and the outcome microvascular invasion, not even male sex being significant.
Table 2 lists the causes of death of all patients who died in the post-transplantation period.There were 14 different causes, the main one being sepsis, with approximately 41%, followed by hypovolemic shock, with approximately 16%.We believe these results were primarily due to liver transplant being a major operation and the use of immunosuppressives postoperatively.
However, significant advances in surgical technique, in the immunosuppressive therapy, and anesthetic management have improved outcomes in the early and late postoperative period 29 .As for the quantitative serum tests, Zhao et al. 22 concluded that the serum concentration of the biomarkers AFP >400mg/L, GGT >130U/L, tumor diameter >8cm, and tumors count >3 were predictors of MVI in patients diagnosed with Multinodular HCC.
In our study, however, we were unable to reproduce these results, as we evaluated tumors larger than 5 centimeters, total bilirubin >1.2mg/dL, platelet count <100x109/L, serum albumin <3.5 or >5.5g/dL, ALT >4U/L or GGT >130U/L, as shown in Table 5.In summary, none of these variables had a significant relationship with microvascular invasion.We also carried out a generalized analysis of the sample, with the patients with one or multiple nodes in the same calculation, although 39% of these were multinodular, very similar to the proportion of 40% in the Yang study sample 30 .
Importantly, other authors 20 had already determined that microvascular invasion is a risk factor for severity and recurrence of HCC, although our study did not contemplate data on hepatocellular carcinoma recurrence.Moreover, when the tumor is close to five centimeters in diameter, the nodule usually begins to lose differentiation and may present microvascular invasion.Based on this, one could assume that the larger the nodule, the worse the patient's prognosis, with a greater probability of recurrence 31,32 .However, we could not confirm this information, as, in addition to not finding a significant relationship between the presence of MVI and mortality, we were also unable to relate the nodule size to MVI, as shown in Table 5.
This can be explained by the possible under-diagnosis of microvascular invasion that existed in the pathology evaluations, especially in older cohorts.
Initially, in Figure 1, the values of the areas under the ROC curves did not show any difference in sensitivity between the MELD and MELD-Na scores for microvascular invasion in patients with HCC.Thus, the curves for both scores were not significant -concluding that both scores could not be used as predictors of MVI.
To date, we have not found any other study with a similar association.The ROC curve (Figure 2) showed a greater sensitivity of MELD-Na over MELD to determine death in patients with HCC, considering the same cutoff point for both.Other authors have shown that the MELD-Na score is more sensitive to determine severity and priority in the transplant queue 18,33 .More than that, we observed a close relationship between the MELD Na values and post-transplant death.From this analysis, we can infer that this score is also useful to assess postoperative prognosis.It is more sensitive than MELD alone, although both scores are statistically significant for this purpose under our analysis.Furthermore, the low specificity of this curve (37.50%) is explained by the large number of deaths resulting from non-hepatic causes, such as multiresistant germs and metabolic disorders.Our study is similar to that of Aranzana et al. 19 , who used the ROC curve for MELD values versus post-transplant survival in seven days, one month, three months, six months, one It is noteworthy that, despite having analyzed a single reference center, our study found data similar to the national profile and to other larger studies 27,30 .
On the other hand, the retrospective analysis generated transplantation at the Hospital das Clínicas (HC) of the State University of Campinas (UNICAMP), in the period of May 2010 to November 2017, due to hepatocellular carcinoma caused by cirrhosis of any etiology.Then, we assessed the possible relationship between the cause of liver disease and other pre-clinical data with the outcomes microvascular invasion and death.In addition, we verified whether the MELD and MELD-Na scores presented differences in sensitivity/specificity to predict death in post-transplant patients..METHODThis is a retrospective study that analyzed the medical records of all patients who underwent liver transplantation due to any disease in the period from May 2010 to November 2017, at the HC UNICAMP, in São Paulo -Brazil.We initially included 125 patients, excluding those whose records did not contain enough information for the present study.

Figure 2 .
Figure 2. ROC curves and areas under the curve for MELD and MELD--Na score values versus death in patients with Hepatocellular Carcinoma.AUC: Area under the curve.

Figure 1 .
Figure 1.ROC curves and areas under the curve for MELD and MELD--Na score values versus microvascular invasion in patients with Hepatocellular Carcinoma.AUC: Area under the curve.
year, and two years, with 1,006 patients.The authors demonstrated that the score could serve as a significant predictor of post-transplant death.Similarly, Xun Luo et al. 34 pointed out the functionality of the MELD-Na score in better predicting both mortality in the waiting list and survival.
fragmentation and loss of information on body mass index (BMI) data, lifestyle habits, and brought uncertainties about the quantification and temporality of alcohol consumption.Thus, we identified the need for new methods and biomolecular parameters to analyze the tumor behavior and to define prognoses and strategies in the pre, intra, and postoperative period and also in the clinical follow-up.

Table 1 .
Clinical impact of the Model for End Liver Disease (MELD) score on the presence of microvascular invasion and on the postoperative outcome in patients undergoing liver transplantation Epidemiological and clinical characteristics of patients undergoing hepatectomy for Hepatocellular Carcinoma.
models in the search for association, using the Odds Ratio for the outcomes "Microvascular invasion" and "Death" due to hepatocellular carcinoma.At first, we carried out a binary logistic regression analysis, and included associations that returned p-values <0.20 in the multivariate logistic regression model.In this, we applied progressive saturation by adding variables, observing the effects on the precision and on the adjustment of the other variables on the definitive analysis.Finally, using the Model End-stage Liver Disease (MELD) score and the variant that includes serum sodium values (MELD-Na), we constructed ROC curves by evaluating the association between the laboratory parameters of liver disease severity (used to calculate the score) and microvascular invasion/death in these patients.We thus aimed to establish the ideal cutoff point on the curve for the outcome as the one that shows the highest values of Sensitivity and Specificity simultaneously.We tabulated all data in Microsoft Excel© spreadsheets and performed the statistical analyzes using the Stata software version 16.0 (StataCorp.2019.Stata Statistical Software: Release 16.College Station, TX: StataCorp LP), with a significance level of 5%.The study is in line with the required ethical precepts, having been previously submitted to the Ethics in Research Committee of UNICAMP, obtaining approval under opinion number 1,377,774.RESULTSTable1shows the individual characteristics of the studied patients.Among the 93 patients, 72 (78.30%) were male, with an average age of 58.53 years (± 0.80 years) and 84 (91.3%)Caucasian.Fortynine (53.26%) of the liver diseases were caused by the Hepatitis C virus, followed by alcoholic liver disease (17.39%).Even so, 65 patients (70.70%) had a history of alcohol intake, 49 (53.26%) were hypertensive, and the tumors were located mainly on the right side (80 -66.67%) of the liver.Tumors had mean size of 3.05cm (± 1.98cm) and most were histological grade 2 (42 -56.25%).Twenty-nine patients (31.52%) had microvascular invasion on histopathological analysis and 44 (47.83%) died, with a time to death of 27.5 days (± 92.5 days), and the main cause was sepsis, totaling 18 (40.91%) of the deaths (Table2).*Meanand standard deviation **Median and Interquartile range

Table 2 .
Causes of death in patients undergoing hepatectomy for Hepatocellular Carcinoma.

Table 3 .
Laboratory characteristics of patients undergoing histopathological analysis after hepatectomy for Hepatocellular Carcinoma.Barbosa Clinical impact of the Model for End Liver Disease (MELD) score on the presence of microvascular invasion and on the postoperative outcome in patients undergoing liver transplantation

Table 4 .
Preclinical factors associated with Microvascular Invasion and death in patients undergoing histopathological evaluation after hepatectomyfor Hepatocellular Carcinoma.

Table 5
dL, GGT >130U/L, and the outcome microvascular invasion.However, no variable remained significant in the multivariate analysis.We identified no associations between the laboratory variables and death.points, with a sensitivity of 90.91% and a specificity of 33.33%.Both scores were significant, with areas under the curves of 0.5727 for the MELD and 0.6032 for the MELD-Na.Deaths occurred on average after 27.5 days, ranging from one day to 53 months.

Table 4
BarbosaClinical impact of the Model for End Liver Disease (MELD) score on the presence of microvascular invasion and on the postoperative outcome in patients undergoing liver transplantation

Table 5 .
Clinical and laboratory factors associated with Microvascular Invasion and death in patients undergoing histopathological evaluation after hepatectomy for Hepatocellular Carcinoma.