Introduction

Hepatocellular carcinoma (HCC) is the most frequent primary liver malignancy [1]. Surgical resection represents one of the best first-line treatments for selected patients. If complete resection can be performed, median survival between 50 and 70 months has been reported [2,3,4]. However, even after curative surgical resection with negative resection margin (R0 resection), intra- or extrahepatic HCC recurrence is frequent, ranging from 40 to 70% [5]. The presence of cirrhosis, tumor >5 cm, positive histological margins, or portal vein invasion have been shown to be risk factors for HCC recurrence [2, 6, 7].

Even though recurrences are frequent, there are no clear international and global algorithms or guidelines for the management of such recurrent lesions, but only recommendations or suggestions by some authors after comparison of one treatment to another. Recommendations are usually based on retrospective studies [8] or on studies originating from Asia where HCC incidence is higher than in Western countries [9, 10]. These recommendations mention that radiofrequency ablation (RFA), chemo-embolization (TACE), and repeat hepatectomy are feasible and safe. Some studies showed that TACE was inferior to repeat hepatectomy and that TACE should be reserved for multilobar involvement [8]. For unifocal recurrent lesion, repeat hepatectomy is usually recommended [11]. For lesion <2 cm, RFA can be an alternative as some studies showed that RFA had similar outcomes as repeat hepatectomy [12,13,14]. Even though several studies have compared the different interventional methods, the real benefits of aggressive treatments in terms of overall survival (OS) are still not clear. The aim of the present study was to assess the OS of treated recurrent HCC after primary surgical resection and to compare it to the OS of non-recurrent cases.

Material and methods

Patients and data collection

From January 1, 1993 to June 30, 2015, all consecutive patients with HCC operated in the Department of Visceral Surgery at the Lausanne University Hospital CHUV (Switzerland) were included. Data on demographics, perioperative details, and postoperative outcomes were retrospectively collected from patient charts. Long-term outcomes were completed by a phone call to the general practitioners if data were missing in the hospital charts. The study was recorded in Research Registry (UIN: researchregistry930) and was approved by the local ethics committee.

Primary hepatocellular carcinoma staging and preoperative workup

The diagnosis of HCC was either based on two imaging modalities typical for HCC with elevated alpha-fetoprotein or on a positive biopsy analyzed by a senior board-certified pathologist in case of atypical images. In case of cirrhosis, the used classifications were the Child-Pugh classification and the Model for End-Stage Liver Disease (MELD) score. Before surgery, the patients benefited from blood tests (complete blood count, liver tests including bilirubin, albumin, and coagulation tests), dosage of alpha-fetoprotein, and radiological assessment with thoraco-abdominal computed tomography (CT) scanner and liver magnetic resonance imaging (MRI). Based on radiological exams, liver volumetry was calculated by the radiologists. In case of estimated remnant liver <40% and cirrhosis, an indocyanine green retention (ICGR) test at 15 min was performed to assess the liver excretion function. In case of cirrhosis, an invasive measure of the portal pressure was also performed. Once resected, the tumor was staged using the TNM classification (7th edition) of the American Joint Committee on Cancer (AJCC). A negative histological margin (R0 resection) was defined as the absence of tumor at >1 mm from the surgical margin.

Treatment strategies of primary HCC

Patients with HCC developed from cirrhosis and within Milan criteria [15] were evaluated for liver transplantation in another institution. Patients with primary HCC not candidates for liver transplantation or on the transplant waiting list (“bridge to transplantation”) with a Child classification grade A or B (i.e., score < 10), a MELD score <9, a remnant liver volume >40% (>30% if no cirrhosis), an ICGR test <15% at 15 min if performed, and a portal pressure <10 mmHg were considered for surgical resection. In case of future liver remnant volume ≤40% (≤30% if no cirrhosis) of the total liver volume, a portal vein embolization of the hemi-liver containing the tumor was undertaken and a new volumetry performed 4 weeks later. All cases of HCC or suspected HCC were discussed in a weekly multidisciplinary tumor board (including radiologists, pathologists, oncologists, hepatologists, and surgeons). Major hepatectomy was defined as liver resection of ≥3 Couinaud’s segments. Postoperative complications were graded according to the Clavien classification [16] and the comprehensive complication index (CCI) [17].

Follow-up

Follow-up after surgery depended on patient age and pathological results. All patients were discussed postoperatively in a multidisciplinary team tumor board. Alpha-fetoprotein levels were measured every 3 months and patients had radiological control (CT or ultrasound) 6 months after the operation. In case of suspicious images, MRI was performed.

Management algorithm for recurrent HCC

Patients with recurrence benefited from a variety of treatments: RFA, TACE, repeat hepatectomy, systemic chemotherapy, radio-embolization, alcoholization, surveillance, or combined treatments. Treatment decisions were based on the type of recurrence (intra- or extrahepatic), comorbidities, and functional liver remnant. All cases were discussed in a multidisciplinary tumor board to tailor the treatment option for each specific patient. In case of recurrence on the resection margin site, repeat hepatectomy was performed if the patient had sufficient remnant liver volume (>40% of total liver volume for cirrhotic livers and >30% of total liver volume for liver without underlying disease), was fit for surgery (ASA score ≤3 and ECOG performance status [18] ≤2), and no major hepatectomy was needed (independently of the lesion size). A recurrence on the resection margin site was defined as new appearance of tumor on the resection margin in case of R0 resection or as tumor evolution on the resection margin after resection with positive histological margin (R1 resection). If the recurrent lesions were multilobar and only intrahepatic, TACE was proposed as the first-line treatment. As second-line treatment, radio-embolization was proposed in case of contraindication to TACE (allergy to chemotherapeutic agents, renal failure, portal vein thrombosis). If TACE and radio-embolization were not possible, alcoholization was performed. If the lesion was unique, <3 cm, and in one lobe, RFA was performed. Liver transplantation was considered in case of recurrence for patients within the Milan criteria [15], but they were not included in this analysis as liver transplantation is not performed in our hospital. In case of extrahepatic disease (lymphatic or organ metastasis), systemic chemotherapy with sorafenib (400 mg twice daily) was proposed. For recurrent cases, time to recurrence, treatment strategies and (overall and 5-year) survival were registered in a database for further analysis.

Statistical analyses

For continuous variables, a Mann-Whitney U test, a Kruskal-Wallis test, or a Student’s t test were used depending on the normality of the distribution and the homogeneity of the variances. Normality was tested using the Kolmogorov-Smirnov and Shapiro-Wilk tests. For categorical variables, a chi-squared test was used. Survival curves were calculated using the Kaplan-Meier method, and the log-rank test was used to compare the survivals. All survivals were calculated from the primary operation date. A p value <0.05 was considered statistically significant. All statistical analyses were performed using GraphPad Prism 5.0© for Mac OS X.

Results

Patients and recurrence characteristics

During the study period, 147 patients were operated for HCC. Among these, 67 (46%) presented a recurrence. Figure 1 shows the flow chart of the recurrent HCC patients included in the study. Demographics and perioperative results of patients with recurrence are presented and compared to patients without recurrence in Table 1. The majority of patients with recurrence were men (52/67, 78%). Forty-five out of 67 HCC developed on cirrhotic livers (67%). Thirty-eight patients had Child A, and 7 Child B cirrhosis. The median MELD score was 8. Among these 45 cases, 25 had an ethylic etiology (56%), 8 a hepatitis B virus infection (18%), 6 a hepatitis C virus infection (13%), 5 a metabolic cause (11%, 4 hemochromatosis and 1 non-alcoholic steatohepatitis), and 1 was from unknown origin (2%). Among the 67 recurrent cases, overall morbidity rate after primary operation was 47% (32/67) with major complications in 19% (13/67). Demographics and surgical results of operated patients with recurrent HCC were similar to operated patients without HCC recurrence.

Fig. 1
figure 1

Flow chart of study patients. HCC hepatocellular carcinomas, RFA radiofrequency ablation, TACE transarterial chemo-embolization

Full size image
Table 1 Demographic characteristics and surgical results of operated hepatocellular carcinoma patients (numbers with percentage and medians with interquartile range)
Full size table

Table 2 presents the characteristics of recurrent HCC patients at time of recurrence. Intra-, extrahepatic recurrences, or both appeared in 57 (85%), 2 (3%), and 8 cases (12%) respectively. The median number of recurrent tumors was 2 (IQR 1–3), and the median size of recurrent tumors was 2 cm (IQR 1–3.5).

Table 2 Characteristics of patients at time of recurrence (numbers with percentage and medians with interquartile range)
Full size table

Treatments of recurrent HCC

Pretreatment decision was based on biopsy in 15 patients and on radiological images only in 52 patients. Figure 1 shows the different initial treatments of recurrent HCC. RFA was the most frequently used treatment for recurrent lesions (18/67). Twenty-six patients benefited from a combination of various treatments (RFA, TACE, repeat hepatectomy, or chemotherapy) over the years. One patient firstly treated by TACE and one firstly treated by RFA underwent afterwards an operation. On the contrary, six patients necessitated a subsequent non-surgical treatment after repeat hepatectomy (3 TACE, 1 RFA, 1 cryoablation, 1 systemic chemotherapy). Eight patients treated by RFA or TACE needed a subsequent RFA or TACE, respectively. Three patients were finally transplanted. Table 3 presents the characteristics and complications of patients with recurrence treated by RFA, TACE, or repeat hepatectomy.

Table 3 Characteristics and complications of operated patients with recurrence treated by radiofrequency (RFA), chemo-embolization (TACE), or repeat hepatectomy (numbers and medians with interquartile range)
Full size table

Recurrence and survival

Median time to recurrence was 11 months after primary surgery. After R0 resection, the median time to recurrence was 12 months (IQR 6–12) compared to 6 months (IQR 4–6) after positive histological margins (p = 0.070). Among the patients who did not survive during follow-up, the median time to death was 30 months in the recurrence group. Patients with recurrence appearing less than 1 year after primary resection had worse median OS compared to patients with recurrence appearing after one year (24 vs. 60 months, p = 0.006).

Median and 5-year OS were lower for the recurrent cases compared to non-recurrent cases (63 vs. 82 months, p = 0.036 and 47 vs. 54%, p = 0.036) (Fig. 2).

Fig. 2
figure 2

Kaplan-Meier survival curves for recurrent and non-recurrent cases (log-rank test, p = 0.036)

Full size image

Among the patients with recurrence, median OS of patients treated by RFA, TACE, or repeat hepatectomy were similar (77, 71, and 84 months, respectively, p = 0.735) (Fig. 3 ). Survival rates of RFA, TACE, or repeat hepatectomy adjusted to MELD scores did not show a difference either (MELD scores <11, p = 0.829). The 4 patients treated by systemic chemotherapy had a median OS of 24 months, and the 17 patients who had palliative care had a median OS of 20 months. Patients treated with chemotherapy and palliative care had lower median OS compared to interventional or surgical treatments (20 vs. 77 months, p < 0.0001).

Fig. 3
figure 3

Kaplan-Meier survival curves for recurrent cases treated by radiofrequency (RFA), chemo-embolization (TACE), and surgery (log-rank test, p = 0.735)

Full size image

Thirty-two patients were treated during the 1993–2004 period and 35 during the 2005–2015 period. There were no differences of interventional and surgical treatments between the two time periods (RFA 8 vs. 10 patients, TACE 6 vs. 10 patients, and repeat hepatectomy 6 vs. 4 patients). Median OS was 60 months during the 1993–2004 period and 61 months during the 2005–2015 period (p = 0.159).

Discussion

This study assessing the long-term outcomes after treatment of recurrent HCC showed that patients with recurrence had worse OS compared to non-recurrent patients. However, in selected patients who are candidates for interventional treatments (RFA, TACE) or repeat hepatectomy, similar survival as non-recurrent cases was observed.

With modern developments of interventional radiology, medical oncology, and surgery, various treatments have been proven to be effective to treat recurrence of operated HCC [19]. Patient selection is important to tailor the treatment to a particular case and multidisciplinary board decision is paramount. There are however no clear international guidelines for these situations. The present study showed that precise selection process and advanced treatments, such as RFA, TACE, and repeat hepatectomy, are valuable strategies in recurrent HCC. Several studies also showed that interventional treatments of the recurrences yield similar OS as non-recurrent cases [5, 20]. Zhang et al. found in their study the same OS between RFA, TACE, and repeat hepatectomy [19]. They additionally found that RFA and repeat hepatectomy were better for late recurrence [19]. Meniconi et al. also showed that survival was longer for patients treated by RFA, repeat hepatectomy, or transplant than by TACE [5]. This effect was not observed in the present study. The type of chemotherapy and of chemo-embolization agent, the use of lipiodol, or the numbers of TACE repetitions could explain this difference of findings.

Patients with recurrences treated by repeat hepatectomy had a median OS of 84 months, which was similar to the OS of patients without recurrence. Of course, patients in that subgroup were well selected (i.e., possibility of total resection of the lesion with negative margins, sufficient estimated remnant liver volume, good performance status). There was also in this subgroup a trend toward less cirrhosis than in the RFA and TACE subgroups (Table 3). Hou et al. found that a repeat hepatectomy for patients with recurrent HCC with microvascular invasion had a median OS of 60 months [21]. They also found that a repeat hepatectomy in recurrent cases without microvascular invasion had better prognosis [21]. A review by Lacaze et al. concluded that repeat hepatectomy and salvage liver transplantation were safe procedures and allowed good long-term survival [11]. In this study, repeat hepatectomy (only minor hepatectomy) was used for recurrent lesion on the surgical margin independently of the lesion size for patients with sufficient remnant liver volume and good general condition. Criteria to perform repeat hepatectomy in case of recurrent HCC considerably vary in the literature.

RFA showed in this cohort the same OS as TACE and repeat hepatectomy (77 vs. 71 vs. 84 months). In the literature, contradictory results are reported in heterogeneous studies, but RFA is accepted as effective in lesions <3 cm, which was the threshold to undertake RFA in this study [13]. Song et al. observed in their study similar survival results between RFA and repeat hepatectomy, corroborating the results of the present study [12]. Chen et al. similarly found in their meta-analysis comparing RFA to repeat hepatectomy that both techniques reached the same OS, and concluded that RFA should be considered as first-intention treatment as it is less invasive [14]. On the contrary, a systematic review of Thomasset et al. concluded that repeat hepatectomy was the most effective treatment for recurrent HCC and that RFA was a safe and effective option for non-surgical candidates [22]. Divergently, in a meta-analysis of five non-randomized studies comparing RFA to repeat hepatectomy, Zhang et al. [23] found no differences in terms of OS with more complications in the surgery group. Finally, Jin et al. showed that TACE was better than RFA for small recurrent HCC lesions [24]. This is in contradiction with the present findings.

Embolization using chemotherapy is now used for different stages of HCC treatment (primary tumor downstaging before surgery, recurrent cases with bilobar involvement, …). In this cohort, TACE was used for recurrent multilobar lesions and had a same median OS as RFA or repeat hepatectomy (71 months). This contrasts with the recent study published by Tabrizian et al. [8], where TACE had a worse OS. A meta-analysis by Wang et al. including 12 studies showed that repeat hepatectomy in case of operated HCC recurrence had better survival than TACE [9]. The heterogeneity was nevertheless high in this meta-analysis, and included studies were of moderate or poor quality.

Salvage liver transplantation, which is also a therapeutic option in case of HCC recurrence [8] was not treated in this article, as only 3 patients finally benefited from this option. A recent study from Mount Sinai Hospital, New York City, showed that salvage transplantation had an intention-to-treat 5-year survival of 44% [8]. The role of salvage liver transplantation remains not clearly defined and should be clarified in the future [11].

In our cohort, follow-up was performed every 6 months during the first 2 years with US, CT, or MRI. As most recurrences appeared less than 1 year after primary operation, it was important to have a radiological follow-up with short time intervals to rapidly detect the recurrence. Several studies showed that recurrences appearing less than a year after the index operation have a worse prognosis due to local tumor invasiveness [20, 25, 26] and intrahepatic metastases [27]. Similar results were found in our study.

The present study has several limitations that must be addressed. Firstly, it is a retrospective cohort. This encompasses all biases inherent to such a study (missing data, charts’ review). Secondly, the group of patients is also largely heterogeneous which renders the analysis complex and definitive conclusions difficult to draw. Moreover, the study period is broad and changes in treatment options have gradually appeared. Two time periods were assessed to show treatment evolution over time and to minimize this bias. There were no differences of treatment in our population between the two periods. This study brings nevertheless new data on recurrences among patients from a European population and reflects clinical day-to-day practice.

Treatment of recurrent HCC is complex, and numerous treatment improvements recently appeared (Table 4). This renders decision-making more difficult. A higher level of standardization is required. In that sense, an international consensus would probably improve the treatment outcomes worldwide.

Table 4 Summary of recent articles (retrospective and prospective studies with minimum 20 patients) published the last 3 years (2014–2016) regarding the outcomes of radiofrequency ablation (RFA), chemo-embolization (TACE), or repeat hepatectomy (RH) for recurrent hepatocellular carcinoma after primary surgical resection
Full size table

In conclusion, recurrence is frequent after HCC resection and negatively impacts OS compared to patients without recurrence. In selected candidates, RFA, TACE, or repeat hepatectomy for HCC recurrence in operated patients offer a similar OS as patients without recurrence. RFA can be performed for unique recurrent lesions <3 cm, TACE for multilobar and intrahepatic only lesions, and repeat hepatectomy for recurrent lesions on the resection margin site in patients with good liver function and enough remnant liver.