Objectives: Microvascular invasion is a well-known risk factor for hepatocellular carcinoma recurrence and mortality after hepatic resection and liver transplant. We sought to determine the clinico-pathological predictive factors associated with microvascular invasion.
Materials and Methods: We studied all patients who had undergone liver transplant because of hepatocellular carcinoma between July 2001 and December 2010 at our institution. Laboratory tests, clinical, and demographic data were obtained. Histopathological hematoxylin and eosin specimens were performed by a single liver pathologist.
Results: During the study, 107 patients had LT because of HCC and they were selected for this investigation: 76 were men (71%) and 31 women (29%) (mean age, 56.8 ± 8.7 y). It was not possible to retrieve histologic samples from 5 patients; therefore, the final studied analysis was 102 individuals. Tumor recurrence rate was 12.9%. One-, three- and five-year overall survivals were 75.0%, 71.4%, and 67.5%. Mitotic index, histologic grade, tumor architecture, alpha-fetoprotein, and tumor fibrosis were associated with microvascular invasion on univariate analysis. Significant independent predictors of microvascular invasion on logistic regression analysis were histologic grade and mitotic index (P < .001; odds ratio, 3.16; 95% confidence interval, 1.525-4.156, and P = .046; odds ratio, 2.56; 95% confidence interval, 1.061-6.451).
Conclusions: Mitotic index and histologic grade are significant predictors of microvascular invasion. No other risk factor was identified in the logistic regression. As both pathological characteristics may be assessed by liver biopsy, these results highlight the importance of discussing pretransplant liver biopsy to access prognosis and define treatment modalities in the setting of liver transplant.
Key words : Primary liver cancer, Tumor recurrence, Prognostic factors
Primary liver cancer is the sixth most common cancer worldwide and the third cause of cancer related mortality; and, hepatocellular carcinoma (HCC) accounts for 85% to 90% of primary liver cancer.1,2
Currently, orthotopic liver transplant (LT) is the treatment of choice in selected patients with HCC and end-stage liver disease, although cancer recurrence is still a major concern owing to its high incidence.3,4 The selection criteria most widely used for indicating LT is the Milan criteria, which takes into account number and size of tumor nodules.5 Pathological data are usually not evaluated before HCC resection or LT, although studies have shown its importance in providing information on recurrence and mortality. Microvascular invasion (MVI) is a well-recognized predictor of worse outcome and post-LT tumor recurrence.6-8 Hepatocellular carcinoma is a highly vascular tumor with a high propensity for vascular invasion, which is usually identified either macroscopically, when the invasion of the vessel is visible on gross examination, or microscopically, when the invasion is visible only on microscopy.9 Given the fact that MVI can only be confirmed as definitely present or absent when the whole tumor is available for histologic examination (after its resection or LT), preoperative methods of assessing the probability of MVI are needed.10 Our aim was to identify clinicopathological risk factors affecting MVI invasion in a group of patients who had undergone LT due to HCC in a single center.
The clinical records of patients who underwent LT at the Hospital das Clínicas, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil, between January 2001 and December 2010 were retrospectively reviewed after the study was approved by the local Ethics Committee. Patients were selected from a group of patients described on an epidemiologic study previously published by our group.11 Data collected included demographic, clinical (cause of the underlying chronic liver disease, history of tobacco use, diabetes mellitus, Child Pugh score, Model of End Stage Liver Disease [MELD] score) and laboratory values (serum alpha-fetoprotein [AFP], creatinine, albumin, total bilirubin and international normalized ratio). Patients with previous diagnoses or those using insulin or oral hypoglycemic agents were assumed to have diabetes mellitus, and those who used tobacco or quit tobacco use 1 year previous to this study were considered smokers.
At our institution LT is performed only in patients fulfilling the Milan criteria (preoperative imaging identifying either a single lesion of ≤ 5 cm or up to 3 lesions ≤ 3 cm).5 Follow-up was defined as the time from transplant to the last visit after LT. Per protocol, visits were conducted on an out-patient basis, and all patients received the same intra- and postoperative care. The main immunosuppressant regimen was based on administering tacrolimus and tapering the dosage of prednisone within 3 months, associated (or not) with mycophenolate mofetil.
The histopathologic assessments were based on the review, by a single liver pathologist, of all hematoxylin and eosin-stained sections. The histologic features evaluated included the presence of fibrosis, steatosis, inflammation (more than 5 inflammatory cell/ high-power field in 10 high-power fields) and necrosis inside the tumor parenchyma. They were reported as being either absent or present. Tumor architecture was assessed for trabecular, pseudoglandular, and solid-growth patterns. Predominant Edmonson and Steiner´s grade was classified and grouped, I and II (as well differentiated) and III and IV (as poorly diffe-rentiated).12 Mitotic activity was counted by reviewing 10 high-power fields and classified as low index (up to 2 mitotic figures in a high-power field) or high index (3 or more mitotic figures in a high-power field) in 10 high-power fields.
The chi-square test or Fisher exact test were used for univariate comparisons between dichotomous variables (sex, cause of cirrhosis, tobacco use, diabetes mellitus, CHILD score, histologic grade, tumor architecture, mitotic index, and presence of steatosis, fibrosis, inflammation and necrosis within the tumor parenchyma), whereas the unpaired t test or Mann-Whitney test were used to compare continues variables as appropriate (i.e., age, creatinine, albumin levels, the MELD score, international normalized ratio, AFP, and total bilirubin). The variables which P values were < .20 in the univariate analysis were included in the multivariate model to assess microscopic vascular invasion. The adequacy of the logistic regression model was accessed by the Hosmer–Lemeshow test. Survival analysis was performed using Kaplan-Meier test. Statistical analyses were performed with SPSS software (SPSS: An IBM Company, version 18.0, IBM Corporation, Armonk, NY, USA).
During the study, 107 patients had LT because of HCC and they were selected for this investigation: 76 were men (71%) and 31 women (29%) (mean age, 56.8 (± 8.7 y). It was not possible to retrieve histologic samples from 5 patients; therefore, the final studied analysis was 102 individuals. The median follow-up was 49.2 months (range, 0-134 mo) and the recurrence rate was 12.9% with a mean time of recurrence of 62.7 months (range, 11.5-127.5 mo). Only vascular invasion was associated with HCC recurrence (P = .05). One-, three- and five-year overall survival was 75%, 71.4%, and 67.5%. The recurrence and survival analyses were performed excluding 14 patients because of early postoperative death, within the first month after LT. The cause of the underlying chronic liver disease was viral hepatitis C in 56.1%, alcohol abuse in 32.7%, viral hepatitis B in 5.6%, cryptogenic cirrhosis in 12.1%, and other causes in 6.2% (2 cases of autoimmune hepatitis, and 1 case of hemochromatosis, primary biliary cirrhosis, the Budd Chiari syndrome, Caroli disease, and nonalcoholic steatohepatitis). At the time of LT, the mean values of the following variables were MELD score 14.7 (range, 6-32), AFP level 255.7 ng/mL (range, 1.2-5592 ng/mL), international normalized ratio 1.59 (range, 0.99-5.6), creatinine 0.99 mg/dL (range, 0.14- 3 mg/dL), albumin 3.1 g/dL (range, 1.1-4.8 g/dL), total bilirubin 2.4 mg/dL (range, 0.2-12.5 mg/dL). Diabetes mellitus was present in 19.8%, and 24.3% of the patients who were smokers at the time of LT. The frequencies of Child Pugh A, B, and C were 25.8%, 51.5%, and 22.7%.
The majority of tumors had a low mitotic index (n= 79/102; 77.5%) and low histologic grade (n = 75/102; 77.3%). The evaluation of tumor parenchyma showed the presence of fibrosis, inflammation, necrosis, and steatosis in 23.5%, 35.6%, 44.1%, and 28.4%.
Microvascular invasion was identified in 38.2% of reviewed samples. Analysis of clinical and histologic factors associated with MVI is shown in Table 1. Mitotic index, histologic grade, AFP, tumor architecture, and tumor fibrosis were selected to be included in the multivariate analysis model. Histologic grade (P < .001; OR, 3.16; 95% confidence interval, 1.525-4.156) and high mitotic index (P = .046; OR 2.56; 95% confidence interval, 1.061-6.451) were independently associated with MVI on logistic regression analysis.
The selection of patients for LT using the Milan criteria was based on tumor size and number, and macroscopic vascular invasion. However, imaging methods can either understate or overstate HCC.13 Thus, is it possible that LT may not be required in some patients. The most important feature not taken into account in the current selection criteria is MVI, which has been shown to be the main prognostic factor independently associated with survival after tumor resection and LT.14-17 Because tumor MVI cannot be determined preoperatively, surrogate markers of MVI are of utmost importance and have been largely investigated. In our study, 38.2% of tumors presented MVI, similarly to what have been previously reported by others.18-19
Two important markers of MVI were identified in multivariate analysis: tumor grade and mitotic index, and both could be recognized in a biopsy specimen before LT. Other pathologic variables were evaluated by the pathologist, including tumor architecture, steatosis, inflammation, fibrosis, and necrosis within the tumor parenchyma, but none showed a positive association with MVI. Parfitt and associates (2007) have already studied tumor architecture, necrosis and inflammation and steatosis within the tumor and found no association with HCC recurrence.20
It has been shown previously that tumor grade is a surrogate marker of MVI.21-24 Furthermore, several studies have demonstrated a direct association of tumor differentiation grade with outcome after treatment.25-27 Colecchia and associates showed that preoperative liver biopsy was a safe and accurate diagnostic tool for tumor grading assessment. Additionally, those authors confirmed a close relation between tumor grade and MVI: that is, the high-grade HCCs were more than twice as likely to have MVI in relation to the low-grade HCCs.28
Mitotic index also has been associated with early tumor recurrence and worse prognosis,29-31 but it was not previously recognized as an MVI predictor, to our knowledge. Schmilovitz-Weiss and associates showed that mitotic activity evaluated by Ki-67 immunostaining was independently associated with mortality.32 Ouchi and associates (2000) found that mitotic index was the main characteristic associated with mortality after HCC resection in the multivariate analyses; therefore, they considered this index a good predictor of poor survival.33 The retrospective nature of the present study might be a limitation, as some clinical information was missed.
We found tumor histologic grade and mitotic index were independently associated with MVI. No other clinical features (cause, sex, age, laboratory tests, MELD score, and Child Pugh) before the histologic assessment predicted MVI. These results are in accordance with previous studies and highlight the importance of discussing pretransplant liver biopsy to access prognosis and define treatment modalities in the setting of LT for HCC.34 Because HCC is a cancer with high recurrence rate and mortality, the use of predictors of MVI may lead to changes in management and follow-up after LT or tumor resection.
Volume : 13
Issue : 5
Pages : 421 - 425
DOI : 10.6002/ect.2015.0045
From the 1Department of Internal Medicine, Faculty of Medicine,
Federal University of Minas Gerais, Belo Horizonte; the 2Alfa
Gastroenterology Institute, University Hospital, Federal University of Minas
Gerais, Belo Horizonte; the 3Department of Pathology and Legal
Medicine, Faculty of Medicine, Federal University of Minas Gerais, Belo
Horizonte; and the 4Department of Surgery, Faculty of Medicine,
Federal University of Minas Gerais, Belo Horizonte
Acknowledgements: The authors have no conflicts of interest to declare. No funding for this study was received.
Corresponding author: Fernanda Maria Farage Osório: R Mantena 51/501 Sion Belo Horizonte - MG: 30310-660, Brazil
Phone: +55 31 9314 2456
Fax: +55 31 3646 3829
Figure 1. Vascular Invasion of a Vessel in High-Power Field
Table 1. Comparative Analysis Between the Clinical and Histologic Characteristics and Microvascular Invasion (univariate analysis)