Neoadjuvant intensity modulated radiotherapy for a single and small (≤5 cm) hepatitis B virus-related hepatocellular carcinoma predicted to have high risks of microvascular invasion: a randomized clinical trial

Background: The presence of microvascular invasion (MVI) significantly impairs postoperative long-term survival of patients with hepatocellular carcinoma (HCC). The role of neoadjuvant radiotherapy (RT) in treating patients with an early-stage HCC predicted to have high risks of MVI remains to be explored. Materials and methods: Consecutive patients with a resectable single and small (≤5 cm) hepatitis B virus-related HCC predicted to have high risks of MVI were randomized 1:1 to receive either neoadjuvant intensity modulated radiation therapy (18 Gy with fractionated doses of 3 Gy) followed by surgery 4 weeks later or upfront surgery. The primary endpoint was disease-free survival (DFS). The secondary outcomes included overall survival (OS), objective response rate, RT-related toxicity and surgical complications. Results: There were 30 patients randomized to each of the two groups. In the neoadjuvant RT group, three patients violated the study protocol, with two having upfront hepatectomy and one radiofrequency ablation after RT. The objective response rate after RT was 25.0% (7/28), but 2 patients suffered from grade 3 liver toxicity. The median follow-up was 68 months (interquartile range, 58–70 months) in the neoadjuvant RT group, and 68 months (interquartile range, 62–75 months) in the upfront surgery group. On intention-to-treat analysis, the median DFS and median OS were not reached in both the 2 arms. The 1-year, 2-year, 3-year and 5-year DFS rates for the neoadjuvant RT group were 86.7%, 76.7%, 60.0% and 56.3%, versus 90.0%, 66.7%, 52.8% and 45.7% in the upfront surgery group (P=0.448), respectively. The corresponding OS rates were 96.7%, 86.7%, 83.3% and 72.7%, versus 100.0%, 93.3%, 79.6% and 60.7% (P = 0.399). Conclusion and relevance: For patients with a resectable single and small hepatitis B virus-related HCC predicted to have high risks of MVI, neoadjuvant RT gave a promising response rate with a mild toxicity. Nevertheless, the neoadjuvant RT yielded similar long-term DFS and OS rates compared with patients who underwent upfront surgery.


Introduction
In 2020, primary liver malignancy was the sixth most common cancer and the second leading cause of cancer-related deaths worldwide [1] .Hepatocellular carcinoma (HCC) accounts for ~75% of all primary liver cancers [2] .Hepatitis B virus (HBV) infection accounts for more than 60% of HCC cases in Asia and Africa, especially in China [3] .Liver resection is a potentially curative treatment which can offer favourable prognosis for early-stage HCC with a single lesion [4] .However, presence of microscopic vascular invasion (MVI) defined as microscopically identified cancer cell nests in portal or hepatic veins in adjacent liver tissues contiguous to the tumour indicates aggressive behaviour of HCC and compromises long-term post-treatment survival of these patients, with a significantly reduced disease-free survival (DFS) at 3 years (relative risk = 1.82) [5][6][7] .Our previous studies suggested that neoadjuvant radiotherapy (RT) prolonged overall survival (OS) and DFS in selected patients with resectable HCC and macrovascular invasion [8,9] .The aim of this prospective randomized controlled trial was to explore whether neoadjuvant RT could provide survival benefits to patients with a single and small ( ≤ 5 cm) HBV-related HCC predicted to have high risks of MVI.

Study design and participants
This randomized, open-label, controlled clinical study was approved by the Ethics Committee of our institute, and its protocol conformed to the standards of the 1975 Declaration of Helsinki.All patients provided written informed consent before participation.The trial also conformed to the 2010 CONSORT guidelines [10] , Supplemental Digital Content 1, http://links.lww.com/JS9/A731 and was registered at the Chinese Clinical Trial Registry.
The inclusion criteria were patients with: (1) age ranging from 20 to 70 years; (2) HCC diagnosed by biopsy or by the noninvasive criteria according to the European Association for the Study of Liver guidelines [11] .(3) hepatitis B surface antigen positivity; (4) a solitary tumour with a maximum diameter less than or equal to 5 cm and was assessed to be resectable using the Criteria in the Appendix, Supplemental Digital Content 2, http://links.lww.com/JS9/A732;(5) preoperative clinical parameters in predicting a high risk of MVI presence on subsequent histopathological study, based on the nomogram developed by Lei et al [12] .This nomogram used seven independent factors to predict presence of MVI with a positive predictive value of 57.9% and a negative predictive value of 83.2%.In this study, we calculated the total score for predicting the risk of MVI using all seven predictive factors from the original nomogram, but with 'tumour number' contributing a fixed score of '0' for all patients due to their having a single tumour.The other six factors, including large tumour diameter, incomplete tumour encapsulation on medical imaging, α-fetoprotein level greater than or equal to 20 ng/ml, platelet count less than 100 × 10 3 / µl, HBV DNA load greater than 10 4 IU/ml, and the presence of the typical 'arterial enhancement and washout' sign on contrast-enhanced MRI, provided variability in the total scores.Patients with a nomogram score greater than or equal to 200 were considered to have high risks of MVI presence and were included in this study.
The exclusion criteria were patients with: (1) a history of other malignancy in the past five years; (2) any previous antitumour treatment for HCC within 1 year; and (3) hepatitis C virus or HIV co-infection.

Sample size estimation
Based on a previous report [13] and the results of subgroup analysis of a large cohort of patients treated from 2003 to 2013 at our institution [14] , the 5-year DFS rate was estimated to be 40% in patients who had a single and small ( ≤5 cm) HCC predicted to have high risks of MVI based on Lei's criteria [12] , and 70% in patients who had no MVI on histopathological examination of resected specimens after surgery [15] .It was also assumed that, after neoadjuvant RT, patients initially at high risk of MVI would achieve a 5-year DFS rate equivalent to that of patients who received upfront surgery and showed no histopathologically identified MVI.Using a two-sided test with 80% power and a significance level of 5%, the minimum sample size was 27 patients for each of the two groups.With 10% of patients being added to allow for subsequent loss to follow-up, 30 patients were determined to be included into each of the two groups.

Randomization
Eligible patients were assigned in a 1:1 ratio to either the neoadjuvant RT or the upfront surgery groups based on a computer-generated randomization code for treatment group allocations without stratification.The allocation was blinded to the researchers who performed the data analyses, but it was not blinded to the patients and clinicians.

Interventions
For patients in the neoadjuvant RT group, intensity modulated radiation therapy (IMRT) was carried out within 1 week after randomization.The gross tumour volume was defined as the tumour volume that was enhanced in the arterial phase.The clinical tumour volume was generated by adding 5-10 mm around the tumour as calculated for the gross tumour volume.The planning target volume was expanded to include a 5-10 mm margin, to compensate for any internal physiologic movements and variations in size, shape and position of clinical tumour volume.The planned total radiation dose for the planning target volume was 18 Gy, delivered in fractions of 3.0 Gy using 6MV X-rays with a linear accelerator (Elekta Synergy), at a rate of 5 fractions per week.Before each treatment session, patients underwent a megavoltage cone-beam computed tomography and were positioned with automated image registration by application of the image guided RT system (Elekta Synergy).The respiratory gating technique was used to reduce the dose delivered to healthy tissues and surrounding organs.Patients undergoing IMRT were put in a supine position with arms above the head and they were instructed to breathe quietly.The RT regimen was in accordance with that used in our previous studies [8,9] .During the course of RT, in case of serious adverse events, which is defined as greater than or equal to grade 3 according to Common Terminology Criteria for Adverse Events (CTCAE) version 5.0 [16] , treatment may be delayed for a maximum of 7 days.If symptoms persisted, further RT or the planned operation was terminated, and patients were recommended to receive other therapies.
For patients in the upfront surgery group, liver resection was carried out within 1 week after randomization.Patients assigned to the neoadjuvant RT group were re-evaluated 4 weeks after completion of RT.If no contraindications were found, hepatectomy was carried out within 1 week.Hepatectomy was performed either anatomically or non-anatomically, based on the liver functional reserve and the possibility of achieving a surgical margin greater than or equal to 1 cm to reduce the possibility of residual MVI at the liver transection plane [17] .Details of the surgical procedure is described in the Appendix.Histopathological examinations were carried out by two independent pathologists using the diagnostic criterion of MVI based on that reported by Roayaie et al [18] .The number and spatial distribution of invaded vessels were utilized to stratify patients with MVI into two groups: the mild MVI group (M1), which comprised patients with 1-5 involved vessels confined within a 1 cm radius from the tumour margin, and the severe MVI group (M2), which included patients with either more than 5 invaded vessels or invaded vessels situated at a distance exceeding 1 cm from the tumour margin [19] .No adjuvant therapy was performed for all enroled patients after surgery, unless an R0 resection had not been performed.

Assessment of outcomes and follow-up
The primary endpoint was DFS, which was defined as the time interval between the randomization to the time disease recurrence was first diagnosed or death from any cause.OS as one of the secondary outcomes was defined as the time from the randomization to death from any cause.Survival outcomes were analyzed in both the intention-to-treat (ITT) and per-protocol (PP) populations.For objective response rate which was used as the other secondary endpoint, it was based on the modified Response Evaluation Criteria in Solid Tumors (mRECIST; version 1.1).RT-related toxicity was evaluated based on the CTCAE version 5.0 [16] .Surgery-related complication was defined as occurrence of any postoperative complication, and graded by the Clavien-Dindo Classification [20] .The protocol of follow-up is described in the Appendix.

Statistical analysis
Variables were compared using the χ 2 test, Fisher's exact test, independent t-test, Mann-Whitney U test or Wilcoxon rank sum test, as appropriate.Survival curves for time-to-event variables were determined by the Kaplan-Meier method, and the log-rank test was used for treatment comparisons.The hazard ratio for survival and its 95% CI were calculated using the Cox proportional hazards regression model.Subgroup analyses on the effect of neoadjuvant RT were performed using clinicopathological variables to include age, sex, serum HBV DNA load, albuminbilirubin grade, α-fetoprotein level, tumour size, as well as the parameters of hepatectomy and pathological examinations.A value of P less than 0.05 was considered statistically significant.All statistical analyses were performed with the Stata 17.0 software (StataCorp).

Patient characteristics
From October 2014 to January 2017, of 182 patients who were treated and assessed for eligibility to be included in this study, 122 were excluded either because of not meeting the inclusion criteria (n = 79) or refusing to participate (n = 43).Finally, 60 patients were randomly assigned in a 1:1 ratio to the neoadjuvant RT group (n = 30) and the upfront surgery group (n = 30; Fig. 1).In the neoadjuvant RT group, three patients violated the study protocol after randomization, with two having upfront hepatectomy and one having radiofrequency ablation after RT.These three patients were also analyzed in the neoadjuvant RT group according to the intention-to-treat principle.
The median scores of the MVI-predicating nomogram were 230 (interquartile range, 219-245) in the neoadjuvant RT group and 232.5 (interquartile range, 219-245) in the upfront surgery group (P = 0.619).Liver cirrhosis was confirmed histopathologically in 63.3% (38/60) of all the included patients, while comparison of liver function status showed no significant difference between the two groups, with the majority of patients having a Child-Pugh score of 5-6 (93.3%) and an albumin-bilirubin grade of 2 (68.3%).The proportions of patients had received anti-viral treatment, defined as a standard course of interferon therapy or the oral administration of anti-viral drugs for a duration of more than 2 months before randomization, were comparable between groups (P = 0.605).(Table 1) In this study, to prevent possible HBV reactivation during treatment, all patients with detectable HBV DNA level ( > 50 IU/ml) not currently undergoing anti-viral therapy were administered with a daily dose of 0.5 mg of entecavir.

Response rate and adverse events of radiotherapy
The objective response rate of RT was 25.0% (7 of 28 patients), with no complete response and progressive disease; seven (25.0%) patients had partial remission and 21 (75.0%) had stable disease.Most of the adverse events after RT were mild and required no treatment.The proportions of patients with greater than or equal to grade 3 adverse events in the neoadjuvant RT group are shown in Table 2.After RT, two patients suffered from grade 3 liver toxicity (two had elevated liver enzymes and one had elevated serum bilirubin) and four patients had grade 3-4 thrombocytopenia.Because of drug treatments of adverse events and for laboratory abnormalities, liver resection was delayed in five patients for 5-7 days.No patients had cancellation of surgery due to the toxicity of RT.

Surgical and pathological information
The proportions of patients who underwent major hepatectomy and anatomical resection, the operation time, volume of blood loss, as well as the probability of intraoperative blood transfusion were comparable between the two groups.Major postoperative complications (Clavien-Dindo classification of grade 3 or above) were observed in one patient in the neoadjuvant RT group.There was no surgery-related mortality in both groups.In the neoadjuvant RT group, the proportion of patients with MVI as confirmed by histopathological examination was 69.0% (20/29, M1: 9, M2:11), compared with 76.7% (23/30, M1: 15, M2: 8) in the upfront surgery group (P = 0.506).(Table 1).

Follow-up data
The median follow-up was 68 months (interquartile range, 58-70 months) for the neoadjuvant RT group and 68 months (interquartile range, 62-75 months) for the upfront surgery  group.At the maximum follow-up of 84 months, 13 patients had developed tumour recurrence and 8 patients had died of HCC in the neoadjuvant RT group, compared with 16 patients who had developed recurrence and 11 who had died of tumour progression in the upfront surgery group.We compared the survival probabilities of patients with severe MVI (M2), mild MVI (M1) and no MVI (M0) using the Cox proportional hazards model.The analysis revealed a hierarchical structure in DFS (P = 0.067) and OS (P = 0.057), with M0 exhibiting the highest survival probability, followed by M1, and then M2, which had the lowest survival probability.(Supplemental Figure A1, Supplemental Digital Content 3, http://links.lww.com/JS9/A733)Patients who received neoadjuvant RT had a higher probability of developing multiple intrahepatic recurrences and distant metastasis versus patients who had upfront surgery (P = 0.041).The details of recurrences are listed in Supplemental Table A1, Supplemental Digital Content 4, http://links.lww.com/JS9/A734.Therapies of recurrences are listed in Supplemental Table A2, Supplemental Digital Content 4, http://links.lww.com/JS9/A734.
Both the DFS and OS rates showed no significant differences between groups of patients stratified by clinicopathological features.The neoadjuvant RT resulted in no significant differences in DFS and OS in patients with histopathologically confirmed severe MVI (M2), mild MVI (M1) and no MVI (M0).(Fig. 3).
The impact of neoadjuvant RT was also evaluated in the subgroups of patients based on their different responses to radiation.In patients who had PR after RT, when compared with those who underwent upfront hepatectomy, neoadjuvant RT was associated with a marginally better DFS (hazard ratio, 0.19; 95% CI, 0.03-1.46)which was approaching a statistical significance (P = 0.076), but there was a similar OS (hazard ratio, 0.30; 95% CI, 0.04-2.31,P = 0.170).(Supplemental Figure A2, Supplemental Digital Content 5, http://links.lww.com/JS9/A735).

Discussion
Liver resection is the established standard of care for patients with Barcelona Clinic Liver Cancer (BCLC) A stage HCC who have a sufficient reserve of liver function [4] .However, the presence of MVI in histopathological examination in resected specimens is correlated with increased incidence of recurrence after hepatectomy [15,18] .Neoadjuvant RT has been shown to be effective in improving postoperative survival in selected patients with HCC  and with macrovascular invasion [8,21] .However, the present study demonstrated no significant differences in long-term DFS and OS in patients with HBV-related HCC who were predicted to have high risks of MVI after neoadjuvant RT followed by hepatectomy compared with patients after upfront hepatectomy both in the ITT or the PP analytic cohorts.Furthermore, on subgroup analyses, differences between postoperative survivals were also not significantly different in patients with histopathologically confirmed MVI using the ITT cohorts for analysis.The adverse events after neoadjuvant RT were generally mild and postoperative complications were comparable between the two groups.
Angiogenesis, being the ability of a tumour to induce new vessel formation to grow and spread, is the fundamental hallmark of cancer, including HCC [22] .MVI occurs when HCC cells have acquired a sufficient capacity to invade blood vessels, commonly portal and hepatic veins [7] .As intrahepatic dissemination of HCC cells is via the portal vein system, the vast majority of HCC recurrence after hepatectomy occurs within the liver remnant.Neoadjuvant therapies having the abilities to reduce tumour volume and destroy microscopic tumour clones can therefore reduce the possibility of tumour recurrence after liver resection for HCC.A number of studies on neoadjuvant local therapies for resectable HCC including Transarterial Chemoembolization (TACE) and Hepatic Arterial Infusion Chemotherapy (HAIC) have been reported.The impact of neoadjuvant TACE on postoperative long-term survival of HCC patients is controversial.There were one prospective randomized trial [23] and two large retrospective cohort studies [24,25] showing no survival benefit by using neoadjuvant TACE.On the other hand, Li et al. [26] reported a 10.5 months survival benefit for patients with a large ( > 10 cm) HCC tumour.Neoadjuvant HAIC before curative liver resection of HCC has been less reported.A retrospective study with a limited sample size revealed an improved event-free survival rate and OS for BCLC stage B/C HCC patients who underwent neoadjuvant HAIC (n = 36) when compared with adjuvant portal vein-infused chemotherapy (n = 38) [27] .
In the old days, RT resulted in suboptimal treatment results in HCC because of low radiation tolerance of liver parenchyma and inability to conform radiation doses to targeted regions [28] .With the development of precision radiation technologies which allow high doses of radiation to be delivered to the tumour while limiting radiation damage to surrounding liver parenchyma, RT has increasingly been applied in treatment of HCC either alone or in combination with other therapies.By inducing tumour necrosis and shrinkage, neoadjuvant RT may be able to eradicate or inhibit the number and extent of MVI, resulting in decrease in recurrence rate.Neoadjuvant RT has been shown in several studies in inducing HCC shrinkage with good safety [8,9,29,30] .In our previous studies on neoadjuvant RT for HCC with major portal vein tumour thrombus, the objective response rates of RT using a daily fractional dose of 3 Gy in 6 consecutive days ranged from 20.7 to 24.0% [8,9] .Only 3.7% patients encountered grade 3 adverse events after neoadjuvant RT, and similar incidences of posthepatectomy surgical complication occurred in the 2 groups of patients with or without neoadjuvant RT.Wu and colleagues carried out a phase 2, single-arm, prospective trial to enrol 38 patients with resectable but centrally located HCC to assess the safety and impact of neoadjuvant IMRT using a treatment regimen of 50-60 Gy in 25-30 fractions over 5-6 weeks.Neoadjuvant RT resulted in a radiological response rate of 42.1% and a major pathological response rate of 34.2%.The resulted grade 3 adverse events only occurred in 3 (7.9%)patients [30] .
The present study employed the same treatment regimen as our previous studies of neoadjuvant RT for HCC with major vascular invasion [8,9] .Although a partial remission rate of 25% was reached in this study, there was no significant difference in postoperative long-term survival between the neoadjuvant RT group and the upfront surgery group.There are several possible reasons to explain why neoadjuvant RT failed to reduce recurrence rate and prolong OS: first, in the previous studies on neoadjuvant RT for HCC with portal vein tumour thrombus, the therapy was found to improve survival outcomes.This improvement was mainly due to the reduction in tumour volume of the portal vein thrombus, which is a known strong and detrimental factor of postoperative survival.The reduction in tumour thrombus volume significantly facilitated subsequent surgical treatment by increasing the rate of en-bloc resection of both the tumour and portal vein tumour thrombus, thus reducing the possibility of residual tumour or spread of tumour cells through the portal vein during surgery [8,9] .However, in this presently reported study, when compared with patients with more advanced HCC, despite the presence of MVI in the HCC patients, the reduction of tumour volume by neoadjuvant RT resulted in significantly less benefit in hepatectomy of early-stage tumours as these tumours have an extremely low possibility of residual lesion after hepatectomy even without neoadjuvant RT.Second, although RT has been shown to be effective in improving survival in HCC patients with MVI in the adjuvant settings [31,32] , the effectiveness and mechanisms involved in the neoadjuvant setting may be different.For instance, partial tumour necrosis of the main tumour induced by irradiation may cause viable tumour cells to become more likely to detach and enter into the bloodstream during hepatectomy [23] .Third, although the response to neoadjuvant RT has been found to positively correlate with more favourable outcomes after surgery, a low total dose of RT (18 Gy) was adopted in our trial to shorten the time interval between neoadjuvant treatment and surgery and to ensure safety to the subsequent hepatectomy, which yielded a relatively low response rate compared with the standard-dose radiation therapy [30,33] .Fourth, a delay in liver resection, coupled with the potential adverse effects of RT on liver parenchyma, could further compromise the surgical outcomes.Nevertheless, our data showed a concordant trend in response to RT and a better DFS approaching a marginal significance, leading to the possibility of using alternative RT regimens or combining RT with systemic drugs such as immune checkpoint inhibitors, to further improve the objective response rate of neoadjuvant therapy and postoperative long-term survival outcomes in patients with earlystage HCC.
This study has limitations.First, this is a single-institution study with a relatively small sample size, and blinding was not possible because of the nature of the intervention.Second, differences in the etiologies of HCC may lead to the different biological characteristics of tumour, thus only patients with HBV-related HCC were included in the MVI-predicating model, as well as in this study.Further research is needed to determine if the results of this study can be generalized to patients with HCC with etiologies other than HBV.Third, due to the current availability, the nomogram employed for predicting a high risk of MVI has relatively high false-positive and false-negative rates, which may limit its use in major clinical decisions.More accurate models are needed to further improve efficacy in selecting patients with a high risk of MVI to receive neoadjuvant therapies.Furthermore, the lack of variability in the 'tumour number' variable, due to all patients in this study having a solitary tumour, might affect the predictive performance of the nomogram for this specific population.This potential limitation should be considered when applying the nomogram in more diverse patient populations.
In conclusion, for patients with a single and small HBV-related HCC who were predicted to have high risks of MVI, although neoadjuvant RT had a promising response rate and a mild toxicity, it yielded comparable long-time disease-free and OSs after partial hepatectomy when compared with those who underwent upfront surgery.

Figure 2 .
Figure 2. Disease-free (A) and overall (B) survival curves for the intensity modulated radiation therapy followed by surgery and upfront surgery groups in the intention-to-treat population.IMRT, intensity modulated radiation therapy.

Table 1
Comparison of clinicopathological characteristics between patients in neoadjuvant radiotherapy group and surgeryalone group.

Table 2
Tumour response and adverse events in neoadjuvant radiotherapy group a .
a Radiotherapy response and safety analyzed in 28 patients who received neoadjuvant radiotherapy.b Patients had decreased platelet count at initial evaluation but no significant decrease ( ≥ 20%) after radiotherapy were excluded.ALT, alanine transaminase; AST, aspartate aminotransferase; CTCAE, Common Terminology Criteria for Adverse Event; mRECIST, modified Response Evaluation Criteria in Solid Tumors.