An E2F7/TSC1-based model for predicting the efficacy of sirolimus after liver transplantation for hepatocellular carcinoma beyond the Milan criteria (2024)

As a radical treatment for hepatocellular carcinoma (HCC), liver transplantation (LT) has become an ideal treatment option for certain HCC patients. For patients beyond the Milan criteria, high HCC recurrence risk is a vital factor affecting the prognosis. Sirolimus, a first-generation mammalian target of rapamycin (mTOR) inhibitor, decreased recurrence rates and prolonged overall survival (OS) compared with calcineurin inhibitors (CNI)-based mTOR inhibitor-free immunosuppression.^[1] However, some patients taking sirolimus after LT regularly do still have HCC recurrence in clinical practice. Thus, predicting the efficacy of sirolimus from the perspective of molecular pathology is expected to break through this bottleneck.

Tuberous sclerosis 1 (TSC1) is a negative regulator of the mTOR complex; so, low expression of TSC1 indicates activation of the mTOR complex. Previously, we found that LT patients with low levels of TSC1 and HCC beyond the Milan criteria had a longer OS and higher recurrence-free survival rates, benefiting from sirolimus-based immunosuppression. In contrast, sirolimus did not improve the prognosis of LT patients beyond the Milan criteria with a high level of TSC1.^[2] E2F transcription factor 7 (E2F7) belongs to the atypical E2F transcription factor family. In in vitro and in vivo models of HCC, E2F7 knockdown significantly promoted cell apoptosis and inhibited the cell proliferation and tumor growth induced by sirolimus. Similarly, we found that LT patients for HCC with low E2F7 expression could benefit from sirolimus in terms of prolonged OS, but patients with high E2F7 expression could not. Based on previous studies, we initially screened the beneficiaries of sirolimus by means of E2F7 and TSC1 single indicators. However, relying solely on E2F7 or TSC1 may have limited clinical guiding value for the use of sirolimus. In addition, E2F7 inhibits mTOR complex 1 by binding to the TSC1 gene promoter to promote TSC1 transcription.^[3] Given that both E2F7 and TSC1 have a negative regulatory effect on the mTOR complex, combining these two indicators to screen beneficiaries of sirolimus may have a greater effect.

This study was conducted in accordance with both the Declarations of Helsinki and Istanbul. This study was approved by the Ethics Committee of the Shulan (Hangzhou) Hospital (No. KY2023071) and First Affiliated Hospital of Zhejiang University (No. 2018-768). Written informed consent was obtained from each participants. In total, 149 LT patients beyond the Milan criteria at two LT centers between 2015 and 2019, were retrospectively included. The inclusion criteria were as follows: (1) patients over 18-year-old; (2) patients who received LT for the first time; (3) survival ≥90 days after transplantation; and (4) no macrovascular invasion. The exclusion criteria were as follows: (1) the patient had combined multiorgan transplantation; (2) the initial time of sirolimus use was within 30 days before death; (3) the initial time of sirolimus use was within 30 days before or after the recurrence of HCC; (4) the patients had a pathological diagnosis of HCC plus intrahepatic cholangiocarcinoma after transplantation; (5) the clinical data were incomplete; and (6) the patient’s primary tumor was derived from extrahepatic organs. The required information obtained from the patients included patient demographics, comorbidities, laboratory tests, radiological data, tumor pathology, treatment before LT, surgical data, and HCC samples.

In all patients, basiliximab (20mg) was regularly administered within 2 h before surgery and on the fourth day after surgery. Methylprednisolone (500mg) was intraoperatively administered. An immunosuppressive regimen based on tacrolimus/cyclosporin A+mycophenolate was implemented in the early postoperative period. In the sirolimus group, sirolimus was usually administered 30–60 days after transplantation. The blood concentration of sirolimus was stable at 4–10ng/mL. At the initiation of sirolimus treatment, the CNI dose was reduced to half, and the CNI was discontinued when the sirolimus target level was reached. Tacrolimus/cyclosporin A was continued in the control group, and the dose was adjusted according to liver function and the blood immunosuppressant concentration. Both groups were treated with mycophenolate.

Tumors isolated from LT patients were processed by formalin fixation and paraffin embedding. Microarrays were established from 149 HCC samples. For TSC1 and E2F7 expression analyses, primary anti-TSC1 (Abcam, USA; ab227594, 1:150) and anti-E2F7 antibodies (Abnova,Taipei, China; Cat# PAB12815, 1:100) were used, respectively. The immunohistochemically stained slides were scanned by an Olympus VS200 scanner (Tokyo, Japan) and evaluated by the staining proportion and intensity, which was called the immune response score (IRS). The IRS was determined by multiplying the staining intensity in four gradations with the percentage of positive cells in four gradations. The gradations were performed as described previously.^[4] To analyze the correlation between E2F7 and TSC1, IRS ≤7 was defined as low, and IRS ≥8 was defined as high. When both E2F7 and TSC1 were expressed at low levels, the combined IRS was 0 points. When either E2F7 or TSC1 was expressed at low levels, the combined IRS was 1 point. When both E2F7 and TSC1 were highly expressed, the combined IRS was 2 points.

The primary endpoint of our study was OS in the transplant patients. Multivariable Cox regression analysis was performed on the variables that achieved P<0.05 in univariable Cox regression analysis. Integrated discrimination improvement (IDI) was used to compare the probability differences in predicting the survival status of beneficiaries of sirolimus screened by E2F7, TSC1, or the combined IRS. Similarly, the area under the receiver operating characteristic curve (AUC) was used to compare the discriminative ability for predicting survival status among the above-mentioned beneficiaries. All statistical tests were two-tailed, with P values <0.05 indicating statistical significance. Statistical analysis was performed using R statistical software (version 4.2.0, https://www.r-project.org).

A total of 205 liver transplants were performed during the study period, of which 149 patients were eligible for clinical trial inclusion [Figure1A]. The average follow-up time for the 149 patients was 34 months (interquartile range, 17.43–45.33 months). These 149 patients were divided into the sirolimus group (n=80) and the control group (n=69). The demographic characteristics, biochemical parameters, and tumor pathology examinations of the two groups are shown in Supplementary Table1, http://links.lww.com/CM9/B978. All preoperative parameters were comparable between the two study arms.

Typical immunohistochemistry staining is shown in Figure1B. A total of 60 patients had low E2F7 expression, and 77 patients had low TSC1 expression. In contrast, 89 patients had high E2F7 expression, and 72 patients had high TSC1 expression. Supplementary Figure1, http://links.lww.com/CM9/B978 indicates that there was no correlation between the expression of E2F7 and TSC1 (Pearson correlation coefficient=0.075, P=0.460). Thus, the combination of E2F7 and TSC1 may have a greater predictive effect on the efficacy of sirolimus.

When patients had low expression levels of E2F7 (n=60) or TSC1 (n=77), sirolimus strikingly improved prognosis (P=0.036 and 0.006), but when they had high expression levels, sirolimus failed to significantly improve prognosis (P=0.330 and 0.670) [Supplementary Figures2A–D, http://links.lww.com/CM9/B978]. According to the combined IRS obtained by analyzing the IHC results, 149 patients were divided into three groups: 0 points (n=38), 1 point (n=61), and 2 points (n=50). Compared with the sirolimus-free group with a combined IRS of 0 and 1 point, sirolimus strikingly prolonged the prognosis of transplant patients with a combined IRS of 0 and 1 point (P=0.004) but not that of patients with a combined IRS of 2 points (P=0.780) [Figure1C, D]. Furthermore, by univariable and multivariable Cox analyses, we determined that tumor differentiation (P=0.004) and administration of sirolimus (P=0.006) were independent prognostic predictors of the 99 patients with a combined IRS of 0 and 1 points [Supplementary Table2, http://links.lww.com/CM9/B978]. Conversely, only tumor differentiation (P=0.004) was an independent prognostic predictor of the 50 patients with a combined IRS of 2 points [Supplementary Table3, http://links.lww.com/CM9/B978].

Compared with E2F7 and TSC1, the beneficiaries of sirolimus screened by the combined IRS increased by 26.2% (39/149) and 14.8% (22/149), respectively, and the groups had similar survival curves. In the 99 patients with a combined IRS of 0 and 1 points, sirolimus had a comparable AUC value at each time point in the 60 patients with low expression of E2F7 and in the 77 patients with low expression of TSC1, which indicated a similar discriminative ability for prognosis [Supplementary Figure3, http://links.lww.com/CM9/B978]. Strikingly, among the 99 patients with a combined IRS of 0 and 1 points, the IDI of sirolimus in precisely predicting the survival status was increased by 8.0% (P=0.005) and 0.2% (P=0.023) compared with low expression of E2F7 (n=60) and TSC1 alone (n=77), respectively. Based on the above results, the use of E2F7 combined with TSC1 to predict the efficacy of sirolimus is a promotion of our previous research results.

At present, mTOR inhibitors, as immunosuppressants, have been extensively applied in the field of organ transplantation, especially in LT for HCC, due to the effect of inhibiting HCC recurrence and protecting kidney function. Nevertheless, a pathway-based approach is broadly thought to become the primary approach to treating cancer; this concept has been potentially challenged by the emergence of acquired drug resistance.^[5] As an inhibitor of the mTOR pathway, sirolimus did not escape the problem of acquired tolerance as well. We also found that there may be discrepancies in the efficacy of immunosuppressive agents for different LT patients with HCC.^[2] Therefore, constructing an efficient prediction model for the efficacy of immunosuppressive agents from the perspective of the molecular typing of HCC to accurately screen the beneficiaries and realize the individualized use of immunosuppressive agents may further improve the prognosis of LT for HCC.

In conclusion, this model innovatively combined the tumor molecular classification and showed accurate prediction performance for the efficacy of sirolimus in LT patients for HCC. Our research results will help clinical decision-making and achieving the individualized and precise application of sirolimus in LT for HCC beyond the Milan criteria.

Funding

The study was supported by grants from the National Natural Science Foundation of China (Nos. 92159202, 82273270, and 32171368) and National Science and Technology Major Project of China (No. 2017ZX10203205).

Conflicts of interest

None.

Footnotes

References