Hepatocellular Carcinoma with Gastrointestinal Involvement: A Systematic Review

In this paper, we aimed to evaluate clinical and imagistic features, and also to provide a diagnostic algorithm for patients presenting with gastrointestinal involvement from hepatocellular carcinoma (HCC). We conducted a systematic search on the PubMed, Scopus and Web of Science databases to identify and collect papers oncases of HCC with gastrointestinal involvement. This search was last updated on 29 April 2022. One hundred and twenty-three articles were included, corresponding to 197 patients. The majority of the patients were male (87.30%), with a mean age of 61.21 years old. The analysis showed large HCCs located mainly in the right hepatic lobe, and highly elevated alfa-fetoprotein (mean = 15,366.18 ng/mL). The most frequent etiological factor was hepatitis B virus (38.57%). Portal vein thrombosis was present in 27.91% of cases. HCC was previously treated in most cases by transarterial chemoembolization (32.99%) and surgical resection (28.93%). Gastrointestinal lesions, developed mainly through direct invasion and hematogenous routes, were predominantly detected in the stomach and duodenum in equal measure—27.91%. Gastrointestinal bleeding was the most common presentation (49.74%). The main diagnostic tools were esophagogastroduodenoscopy (EGD) and computed tomography. The mean survival time was 7.30 months. Gastrointestinal involvement in HCC should be included in the differential diagnosis of patients with underlying HCC and gastrointestinal manifestations or pathological findings in EGD.


Introduction
Hepatocellular carcinoma (HCC) is the most common primary malignant tumor of the liver and the sixth most common cancer, according to GLOBOCAN 2020 data. Although both incidence and mortality rates declined in many high-risk areas, many patients have already reached an advanced stage at diagnosis, resulting in 830,000 deaths worldwide [1].
Hepatocellular carcinoma usually disseminates to the liver [2]. Although less common, in 30-50% of the cases, HCC can have extrahepatic spread. The most frequent areas are the lungs, followed by lymph nodes and bones [3][4][5]. Involvement of the gastrointestinal (GI) tract is a rare event, with a reported incidence of 0.5-2% of all HCCs. Higher rates of 4-12% have been recorded in autopsy cases [6,7].
Due to non-specific symptomatology or endoscopic features, GI involvement by HCC is underdiagnosed premortem [8]. The available data about this condition are provided mainly by case reports and only a few literature reviews. In 2004, Fujii, K. et al. reviewed the characteristics of 29 HCC patients with GI tract invasion [9]. Later on, in 2011, Lin, T.L. et al. assessed the course of disease and survival in 44 patients reported in the English literature with direct invasion of the GI tract by HCC [10]. In the same year, Kato, Y. et al. (2011) investigated the role of surgical treatment based on 18 cases from the literature (English literature and Japanese literature with English abstract), including his reported case [11]. In 2018, Harada, J. et al. also reviewed the clinical characteristics of esophageal metastases from HCC [12]. Recently, Yu [13,14]. However, these studies included a limited number of cases and were focused on a specific route of metastasis, or the involvement of a particular segment of the GI tract. Additional cases have been reported since these previous publications.
The HCC survival rate has increased over the last three decades and we expect it to further increase as a result of improvements in therapy and early diagnosis [15]. For that reason, we also expect to see a more significant number of patients with atypical complications in our clinical practice [16]. An early diagnosis of GI involvement from HCC is a challenge for clinicians, and raising awareness of this issue is a crucial step toward it. Our systematic review complements previous studies and gives a bigger picture of the main clinical and imagistic characteristics of GI involvement from HCC. We also propose an algorithm diagnosis that would serve clinicians in making a rapid diagnosis.

Materials and Methods
A systematic review of case studies of gastrointestinal involvement from hepatocellular carcinoma was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Table S1) [17].

Data Sources
The systematic search was performed in the PubMed, Scopus and Web of Science databases up to November 2021, with the last search update on 29 April. A search in the Scopus and Web of Science database was also performed to avoid missing relevant articles up to 29 April. Randomly discovered searches, mainly from the manual search of references, were also included.

Inclusion and Exclusion Criteria
Our search strategy was developed on PICO (Patient/Population, Intervention, Comparison, Outcome) elements as follows: population: patients with an imagistic or histopathological diagnosis of hepatocellular carcinoma involving the gastrointestinal tract; intervention: none; comparison: none; outcomes: case studies that reported clinical presentation, diagnosis or survival of HCC with gastrointestinal involvement. All published case reports, case series or retrospective observational studies on the PubMed database concerning GI involvement from hepatocellular carcinoma were considered eligible for inclusion. We extended our search to editorial material that provided enough data for retrieval and analysis. No restriction for the time of publication was fixed. Only articles that addressed the specific clinical topic were selected for the examination. We excluded articles written in languages other than English, those not focused on the main issue, and papers not available as full text or papers with insufficient data to support the analysis.

Search Strategy
The search strategy included a combination of the following terms connected with the Boolean operators "AND" and "OR": "hepatocellular carcinoma", "esophagus", "stomach", "gastric", "duodenum", "jejunum", "ileum", "intestine", "colon", "rectum", "invasion", "metastasis". No filters were added. The search strategy is reported as Supplementary Materials (Table S2). In the first stage, duplicate references were removed from our results using EndNote 20 (Clarivate Analytics, Chandler, AZ, USA). Further, tiles/± abstracts of the records found in the first stage were screened by two reviewers, and reports that were not on the main subject were excluded. Cohen's Kappa coefficient was calculated, obtaining a satisfactory value of 0.94. The next step was to eliminate non-English articles. The full text of the articles was retrieved when available and further analyzed. Articles containing insufficient data and those not related to the main topic were excluded.
Further, we composed a data-extracting sheet in Microsoft Excel (Microsoft Office Professional Plus 2019), Microsoft ® Excel ® 2019 MSO (Version 2111 Build 16.0.14701.20204) 32-bit. One member of our team performed the initial data extraction and it was checked for reliability by a second member.
The following data were extracted: additional bibliographic information, including name of the first author, year of publication and type of study; number of patients reported in the paper; patient's demographics, such as gender and age; etiology; location and size of HCC; previous treatment for HCC; AFP value (ng/mL); presence of portal vein thrombosis; the involved segment of the gastrointestinal tract; clinical presentation; involvement route; imagistic methods used for diagnosis of GI lesions; endoscopic features; methods used to obtain histological specimens and whether or not immunohistochemistry was used, as well as the survival time. The extracted data were further processed. Descriptive statistics (mean, standard deviation, percentages, minimum, maximum) were used to summarize the analyzed data.
Any discrepancy between reviewers was discussed, and a final consensus was reached.
In order to assess the risk of bias in our systematic review of case reports and case series, we used the Joanna Briggs Critical Appraisal Checklist for case reports and for case series (Tables S3 and S4) [18]. Figure 1 illustrates the PRISMA flow diagram of our search strategy. We initially identified 9474 record titles, from which we excluded 2727 duplicates. Further, from the remaining 6747 records, only 195 were sought for retrieval. Forty-seven articles were non-English publications. In the last stage, from the 148 reports considered eligible, one was an E-poster, 26 did not have full-text availability, three were not on discussed topic and five did not have sufficient detailed data to support the analysis (Table S5). We identified ten additional publications on manual search. One hundred twenty-three articles concerning 197 patients were included in the final analysis (Tables 1 and 2).

Patient Characteristics
We included 197 cases, with a mean age of 61.21 (standard deviation = 11.66), ranging from 22-86. The majority of patients were male (n = 172; 87.30%), with a ratio of man: woman of 6.88.

Etiology
Data concerning the etiology of liver disease in patients with HCC were described in 158 cases (Table 3). Hepatitis B virus (HBV) was incriminated in most patients (38.57%), followed by hepatitis C virus (HCV) (17.76%) and alcohol (7.61%). Coinfection with HBV and HCV was reported in 2.03% of cases, and only 0.5% of patients were identified with HBV, HCV, and hepatitis D virus (HVD) coinfection. Autoimmune etiology and nonalcoholic fatty liver disease (NAFLD) were suggested in one patient. In 19 cases, the underlying cause was unknown (9.64%).

Clinical Findings in the Study Population
A summary of the clinical characteristics of the study population is described in Table 4. Most HCCs were bulky masses, with an average tumor size of 8.66 cm (n = 92 hepatic nodules). Liver tumor localization was described in 158 patients. Four patients did not have any tumor recurrence at the moment of diagnosis. Hepatocellular carcinoma was located as follows: right hepatic lobe (31.47%), left hepatic lobe (21.82%), both hepatic lobes (19.28%), caudate lobe (2.03%), hepatic hilum (1.01%), peritoneum (1.01%), left hepatic lobe and caudate lobe (0.50%), right hepatic lobe and caudate lobe (0.50%) and lymph nodes (0.50%). Portal vein thrombosis was found in 27.91% (55/197) of the evaluated cases. Regarding the tumoral markers, the mean value of serum AFP was 15,366.18 ng/mL (n = 112 available data, 14 patients were reported as having a normal value).

Involved GI Site and Presumed Mode of Involvement
The most commonly involved sites in the gastrointestinal tract were the stomach (55/197; 27.91%) and duodenum (55/197; 27.91%,), followed by colon (32/197, 16.24%), esophagus (18/197; 8.92%), jejunum and ileum (14/197; 9.13%,) and rectum (3/197; 1.52%). Synchronous localization occurred in 9.64% of patients. The sites with concomitant involvement were: stomach and esophagus, stomach and duodenum, stomach and colon, stomach and small bowel, duodenum and colon and rectum and colon. Our study suggested that in most cases, GI involvement occurred through direct invasion (87/197; 44.16%). HCC metastasized through the hematogeneous route in 31.97% of situations (63/197). Translymphatic dissemination was reported for 3.04% (6/197); meanwhile, peritoneal spreading was found in 3.55% of patients (7/197). We also detected three cases of iatrogenically induced metastases. Moreover, both direct invasion and hematogenous spread were considered in two patients with concomitant duodenal and colon involvement and esophagus and stomach involvement, respectively. Details of the segments of the GI tract involved and the routes of involvement are listed in Table 6.

Prognosis of Gastrointestinal Involvement in Patients with HCC
Prognosis of patients with hepatocellular carcinoma and GI tract involvement (n = 158 available data; from which 12 were lost to follow-up) was dismal, with an average survival of 7.30 months. In 3.04% of cases (6/197), the diagnosis was made post-mortem at autopsy, and 1.01% (2/197) survived for less than 24 h. In the present study, only 14.72% (29/197) were still alive at the moment of the last follow-up.

Discussion
Hepatocellular carcinoma yields high recurrence rates, even after radical resection. Liver transplantation is the best treatment method because it also cures the underlying liver disease, but it is not broadly applicable [132]. However, other available therapeutic approaches, such as locoregional therapies, have been developed with the purpose of increasing the survival rate in patients with unresectable HCC [133]. Improvements in the survival period are associated with a higher risk of developing extrahepatic metastases. Among them, gastrointestinal involvement from hepatocellular carcinoma is rare [118].
In our study, both stomach and duodenum were the areas of the GI tract most frequently affected. In reviews reported in literature, the stomach was most commonly involved, followed by duodenum [10]. On the other side, esophagus metastases were very uncommon in our analysis, accounting for less than 9% of cases. Sites of the lower part of the gastrointestinal tract, such as the colon, jejunum, ileum and rectum were also less affected. In exceptional cases, a liver tumor can simultaneously involve more than one segment of the GI tract [7,51,58,59,65,66,80].
As in most reports, direct invasion was the predominant spread pattern. Factors favoring GI involvement were growth mode, size and localization of hepatic tumors [14]. Due to the the anatomical relationship between the liver and GI tract segments, HCCs localized in the right hepatic lobe tend to invade the duodenum, and those located in the left lobe usually involve the stomach [84]. The role of TACE is controversial. TACE can induce tumor adherence to the liver capsule and GI tract through necrosis and inflammation. On the other hand, HCC was diagnosed concomitant with the GI invasion in many cases, and these patients had not received any previous treatment [14]. In our analysis, TACE was previously performed in 32.99% of cases, and 26.90% did not receive any treatment.
The hematogenous route was the second most frequent path. Although it can be detected in liver cirrhosis, reverse portal flow is more frequently observed in primary hepatocellular carcinoma due to arteriovenous communications and arterial neovascularization [134]. Tumor emboli can be disseminated from the liver to the gastrointestinal tract by the hepatofugal portal flow [131]. Portal vein thrombosis is also a significant contributing element that can exacerbate the reversal of the flow [119]. This aspect is supported by the fact that 27.91% of the assessed cases in our study presented portal vein thrombosis. It is hypothesized that endoscopic therapy of esophageal varices, in particular esophageal band ligation, can promote the development of esophageal metastasis by trapping the tumoral thrombi [31]. Hiraoka, T. et al. (1986) presented two cases of hepatocellular carcinoma with invasion of the portal vein branches, in which microscopic tumoral thrombi were found in sclerosed esophageal varices [20]. Kume, K. et al. also presented a case of HCC metastases developed at the place of variceal band ligation [21].
In our review of the literature, GI tract involvement in HCC was also reported to develop after liver transplant [25,28,29,32,41,55,121]. The risk of developing uncommon metastases of HCC, including gastrointestinal metastases, can increase after a liver transplant due to a delicate physiological state or to the administration of immunosuppressive agents [32].
In exceptional cases, dissemination of tumor cells through needle track following endoscopic ultrasound-fine needle aspiration (EUS-FNA) performed for the confirmation of HCC has also been described [61]. Moreover, periampullary metastasis from HCC has been reported after biliary interventions in a patient with HCC invaded in the biliary tract [84].
Gastrointestinal bleeding, either frank or occult, was the most common presenting feature among the studied cases. Similarly to our results, in an analysis of 30 cases with direct GI involvement from HCC, reported in the English literature, Korkolis, D.P. et al. (2009) also concluded that gastrointestinal bleeding was the main clinical presentation [46]. Besides GI bleeding, the spectrum of clinical manifestations was vast in our study results. It included abdominal pain, palpable mass, chronic anemia, dysphagia, fatigue, weight loss, nausea, vomiting, diarrhea and gastric outlet obstruction.

Algorithm of Diagnosis
On the basis of data found in the literature and the results of our research, we propose a diagnostic algorithm (Figure 2). The diagnosis can be extremely challenging and clinicians must be aware of this condition, as early detection and prompt treatment are crucial for a better prognosis. In patients with a known clinical context of cirrhosis and hepatocellular carcinoma, with new-onset gastrointestinal symptoms, especially GI bleeding, gastrointestinal involvement from HCC should be considered as a possible etiology if other common causes are excluded [47,71]. Clinical examination plays a key role in identifying an abdominal mass [100]. In the next stage, risk-factor assessment should be conducted: tumor size, location, growth pattern, presence of portal vein thrombosis and previous locoregional therapies for the primary liver tumor, or even liver transplant and endoscopic therapy for esophageal varices [14,31,32,119]. Endoscopic examination is the standard gold method for identifying GI lesions [7]. As evidenced by the various aspects described, endoscopic features are not specific and can pose differential diagnosis problems. However, the following aspects should draw suspicion in a patient with HCC: polypoid mass, a submucosal tumor/extrinsic compression, ulcerative lesions or even the presence of a fistula [8,48,50,63,80]. The histopathologic examination is mandatory for a certain diagnosis. In some situations where there is uncertainty, investigations should include immunohistochemical tests [7,114,126]. Hepatocyte paraffin-1 (Hep par-1), glypican-3 (GPC-3), arginase-1 and polyclonal carcinoembryonicantigen (pCEA) effectively differentiate GI metastases of HCC from other types of tumors [126]. EUS and EUS-FNA are the alternative diagnostic methods for GI submucosal lesions or for when endoscopy fails to identify the tumor [33]. Radiological investigations are an excellent guidance modality. CT can describe the localization, size and extension of primary liver tumor, the status of the portal vein and lymph nodes, the site of invasion, the contiguity of HCC with the GI tract lesions and the severity of underlying liver cirrhosis, and can also exclude other metastases [8,118]. GI metastases of HCC usually display hyperenhancement in the arterial phase on CT scans, similarly to liver tumors [8]. FDG-PET/CT and angiography can complete the diagnostic workup [6,7,29,32,33,71].
Based on our reviewed articles, the mean survival was 7.30 months. Fujii, K. et al. (2004) evaluated median survival in 29 patients with HCC invading the GI tract. The estimated median survival time was 1.2 months for patients who received supportive treatment, and three months for nonsurgical treatment; meanwhile, patients treated with curative surgery showed an average survival of 9.7 months [9].
Although it may improve the clinical approach of cases with HCC and GI involvement, our study has a number of limitations. Subjectivity in data interpretation, lack of a large control group and the fact that only English-language articles were included are just some of these limitations, and may have influenced the final results.

Conclusions
In conclusion, to our knowledge, we are reporting the most extensive systematic review of case reports to date on the involvement of the gastrointestinal tract in HCC. Gastrointestinal involvement in HCC could be included in the differential diagnosis of patients with underlying HCC and gastrointestinal manifestations or pathological findings in EGD.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/diagnostics12051270/s1. Table S1: PRISMA 2020 checklist. Table S2: Complete search strategy. Table S3: Methodological quality of included case reports using JBI Critical Appraisal Checklist for case reports. Table S4: Methodological quality of included case reports using JBI Critical Appraisal Checklist for case series. Table S5: Relevant studies excluded from the systematic review.