The Neoangiogenic Transcriptomic Signature Impacts Hepatocellular Carcinoma Prognosis and Can Be Triggered by Transarterial Chemoembolization Treatment.

Citation: Critelli, R.M.; Casari, F.; Borghi, A.; Serino, G.; Caporali, C.; Magistri, P.; Pecchi, A.; Shahini, E.; Milosa, F.; Di Marco, L.; et al. The Neoangiogenic Transcriptomic Signature Impacts Hepatocellular Carcinoma Prognosis and Can Be Triggered by Transarterial Chemoembolization Treatment. Cancers 2024,16, 3549. https:// doi.org/10.3390/cancers16203549 Academic Editor: Christoph Reissfelder Received: 26 August 2024 Revised: 5 October 2024 Accepted: 15 October 2024 Published: 21 October 2024 Copyright: © 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). cancers Article The Neoangiogenic Transcriptomic Signature Impacts Hepatocellular Carcinoma Prognosis and Can Be Triggered by Transarterial Chemoembolization Treatment Rosina Maria Critelli 1, Federico Casari 2, Alberto Borghi 3, Grazia Serino 4, Cristian Caporali 2, Paolo Magistri 5, Annarita Pecchi 2, Endrit Shahini 4, Fabiola Milosa 1, Lorenza Di Marco 6, Alessandra Pivetti 1, Simone Lasagni 1, Filippo Schepis 7, Nicola De Maria 1, Francesco Dituri 4, María Luz Martínez-Chantar 8,9 , Fabrizio Di Benedetto 5, Gianluigi Giannelli 4and Erica Villa 7,* 1Gastroenterology Unit, CHIMOMO Department, University of Modena and Reggio Emilia, 41124 Modena, Italy; [email protected] (R.M.C.); [email protected] (F.M.); [email protected] (A.P.); [email protected] (S.L.); [email protected] (N.D.M.) 2Radiology, Azienda Ospedaliero-Universitaria di Modena, University of Modena and Reggio Emilia, 41125 Modena, Italy; [email protected] (F.C.); [email protected] (C.C.); [email protected] (A.P.) 3Internal Medicine, Ospedale di Faenza, 48018 Faenza, Italy; [email protected] 4National Institute of Gastroenterology “IRCCS Saverio de Bellis”, Research Hospital, 70013 Castellana Grotte, Italy; [email protected] (G.S.); [email protected] (E.S.); [email protected] (F.D.); [email protected] (G.G.) 5HPB Surgery and Liver Transplant Unit, Azienda Ospedaliero-Universitaria di Modena, University of Modena and Reggio Emilia, 41125 Modena, Italy 6Clinical and Experimental Medicine PhD Program, 41125 Modena, Italy; lor[email protected] 7M.E.C. Dipartimental Unit, University of Modena and Reggio Emilia, 41125 Modena, Italy; [email protected] 8Liver Disease Laboratory, Centre for Cooperative Research in Biosciences (CIC bioGUNE), Basque Research and Technology Alliance (BRTA), Bizkaia Technology Park, Building 801A, 48160 Derio, Spain; [email protected] 9Centro de Investigacion Biomedica en Red de Enfermedades Hepaticas y Digestivas (CIBERehd), 28200 Madrid, Spain *Correspondence: [email protected]; Tel.: +39-0594225308; Fax: +39-0594222624 Simple Summary: Therapeutic response and survival outcomes in hepatocellular carcinoma (HCC) patients remain unsatisfactory, with a 5-year survival rate of less than 25%. The lack of molecular analysis of HCC tissue hinders the identification of precise predictors of disease progression and treatment outcomes. Analyzing the hepatic neoangiogenic transcriptomic signature can predict the biological aggressiveness of HCC and its resistance to therapies, offering a valuable diagnostic and prognostic tool that can significantly enhance the management of HCC. Abstract: Background/Objectives: We evaluated the relationship between the neoangiogenic transcriptomic signature (nTS) and clinical symptoms, treatment outcomes, and survival in hepatocellular carcinoma (HCC) patients. Methods: This study prospectively followed 328 patients in the derivation and 256 in the validation cohort (with a median follow-up of 31 and 22 months, respectively). The nTS was associated with disease presentation, treatments administered, and overall survival rates. Additionally, this study investigated how multiple treatments influenced changes in nTS status and alterations in microRNA expression. Results: The nTS was identified in 27.4% of patients, linked to aggressive features like multifocality and elevated alpha-fetoprotein (AFP), a pattern consistent with that of the validation cohort. Most patients in both cohorts received treatment for HCC. nTS+ patients had limited access to, and benefited less from, liver transplantation or radiofrequency ablation (RFA) compared to nTS − patients. By the end, 78.9% had died, with nTS − patients showing better median survival and response to treatments than their nTS+ counterparts, who had lower survival across all treatment types. Among those who received transarterial chemoembolization (TACE), 31.2% (21/80 patients after the initial treatment and another four following a second TACE) transitioned Cancers 2024,16, 3549. https://doi.org/10.3390/cancers16203549 https://www.mdpi.com/journal/cancers Cancers 2024,16, 3549 2 of 16 from an nTS − to an nTS+ status. This shift was associated with lower survival and alterations in microRNA expressions related to oncogenic pathways. Conclusions: The nTS markedly influences treatment eligibility and survival in patients with HCC. Notably, the nTS can develop after repeated TACE procedures, significantly impacting patient survival and altering oncogenic microRNA expression patterns. These findings highlight the critical role of the nTS in guiding treatment decisions and prognostication in HCC management. Keywords: neoangiogenic transcriptomic signature; hypoxia; microRNA expression; survival 1. Introduction Hepatocellular carcinoma (HCC) remains a global health challenge despite advances in screening and treatment, causing over 700,000 deaths annually with a five-year survival rate under 10% [ 1 ]. HCC commonly arises as a complication in patients with long-term liver cirrhosis, affecting roughly a third of this population [ 2 ]. Prognosis hinges on the stage of liver disease, tumor traits, and treatment [1]. Until now, prognostic evaluations and treatment decisions have been based on staging systems like the widely adopted Barcelona Clinic Liver Cancer (BCLC) classification [ 3 ], supported by European (EASL) [ 4 ] and American (AASLD) guidelines [ 5 ]. However, BCLC’s predictive accuracy at the individual level has been questioned. The main drawbacks include imprecise stratification of patients into the BCLC-B subclass and a certain degree of rigidity in stage-specific therapeutic choices, leading to poor adherence to its therapeutic indications in real-life clinical practice. As a result, its use in everyday clinical practice has become more limited [ 6 – 9 ]. Furthermore, most staging systems, including BCLC, consider only initial tumor characteristics, overlooking tumor growth and the emergence of new lesions [10]. There is a growing consensus that integrating clinical staging with biological markers of the tumor could refine prognoses and guide treatment strategies [ 11 , 12 ]. Nonetheless, the complexity and limited added value of many identified molecular signatures have hindered their adoption in clinical settings. Additionally, most signatures are based on retrospective data from a small fraction of HCC patients eligible for resection, limiting their applicability [13]. One tumor characteristic linked to prognosis is the growth rate [ 14 – 16 ]. Fast-growing HCCs, constituting 20–30% of cases, are notably aggressive and less responsive to treatments [ 15 – 17 ]. In our previous work in 2016 [ 15 ], we identified a five-gene neoangiogenic transcriptomic signature (nTS) through an extensive microarray study. This signature includes angiopoietin-2, Delta-like canonical notch ligand 4, neuropilin and Tolloid-like 2, endothelial cell-specific molecule 1, and nuclear receptor subfamily 4 group A member 1, all associated with neoangiogenesis. HCCs in patients with the nTS exhibited not only an extremely fast growth rate but also a distinctly immunosuppressed microenvironment, evidenced by the local upregulation of PD1 (Programmed cell death protein 1) and PD-L1 (Programmed Death Ligand 1) [ 18 ]. Additionally, these HCCs showed prominent epithelial– mesenchymal transition and clear activation of TGF β 1 signaling. Overall, the nTS was linked with aggressive HCCs and poor prognosis [ 18 ]. In this study, we aim to evaluate how this signature and its changes after treatment affect patient outcomes and survival in two prospectively enrolled HCC cohorts. 2. Patients and Methods 2.1. Patients From January 2010 to January 2017, the Gastroenterology Unit at Azienda OspedalieroUniversitaria, Modena, prospectively enrolled newly diagnosed treatment-naïve HCC patients detected via biannual ultrasound surveillance. Treatment followed global guidelines: early-stage HCCs underwent surgical resection or radiofrequency ablation (RFA), interme- Cancers 2024,16, 3549 3 of 16 diate stages received transarterial chemoembolization (TACE), and advanced stages were treated with systemic therapies like sorafenib. Upon sorafenib cessation, options included regorafenib or participation in clinical trials, such as milciclib (n= 2) or nivolumab ( n= 2 ). Best supportive care (BSC) was reserved for those with advanced liver disease or poor performance status (PS) [Child–Pugh (CP) B 8/9 and/or PS > 1]. Liver transplant eligibility and MELD score adjustments followed Italian transplant guidelines [ 19 ]. Tumor response was measured according to the modified RECIST criteria [ 20 ]. Response was defined after six months of treatment as stable disease (SD) or partial response (PR) or complete response (CR). Post-treatment imaging with contrast-enhanced Computed Tomography or Magnetic Resonance Imaging was conducted at one month and every three months for two years post complete treatment. Recurrences triggered stage-appropriate retreatment. Those without recurrence for over a year were monitored with semi-annual ultrasounds. We confirmed our findings with a validation cohort, also prospectively enrolled, in our unit from 2018–2021, due to the unavailability of external cohorts with suitable baseline biopsies for HCC. This group underwent identical diagnostic procedures to the derivation cohort. Conducted in line with the Declaration of Helsinki and clinical trial practices, the protocol was approved by an ethics committee, with all participants providing informed consent (IRB10/08_CE_UniRer; ClinicalTrials ID: NCT01657695). 2.2. Neoangiogenic Transcriptomic Signature After written informed consent, all patients underwent an ultrasound-guided biopsy for both histological examination and transcriptomic analysis to determine the presence of the neoangiogenic transcriptomic signature (nTS) [ 15 ]. In patients experiencing recurrence, after collection of informed consent, a second biopsy or a third biopsy was obtained. In these cases, histological examination and transcriptomic analysis were also performed. Tumor and non-tumor tissue were collected in ice-cold RNA-later and processed within 24 h ([ 15 ] and Supplementary Materials). In instances of repeated treatments, another biopsy was performed to assess any changes in the nTS. The physicians responsible for patient care were blinded to the nTS results, ensuring that this information did not influence the therapeutic decision-making process. 2.3. microRNA Analysis We utilized a segment of the ultrasound-guided biopsy for miRNA expression analysis. This was performed using the miRCURY LNA RT Kit from Qiagen s.r.l. (Milan, Italy) ([ 18 ] and Supplementary Materials). The following miRNAs were evaluated: miR-221-3p and miR-222-3p (involved in promotion of proliferation, migration, invasion, and hypoxiadriven angiogenesis) [ 21 – 23 ]; mir-15b-5p and mir-16-5p (inhibition of apoptosis, growth, and upregulation in the hypoxic environment) [ 24 ]; miR-30a-5p (inhibition of proliferation, invasion, tumor growth) [ 25 – 27 ]; miR-30d-5p (inhibition of autophagy) [ 28 , 29 ]; miR-145-5p (negative regulation of cell proliferation) [ 30 , 31 ]; mir-122-5p (cell cycle arrest, EMT, and apoptosis) [32,33]; and miR-210 (hypoxia) [34–37]. 2.4. Statistical and Bioinformatic Analysis Continuous variables were presented as means ± standard deviations (SDs), and comparisons were made using Student’s t-test. Categorical variables were expressed as frequencies of patients with and without the nTS and were compared using Pearson’s χ2 test or Fisher’s exact test, depending on group size. The Mann–Whitney U test was employed for comparisons when data were assumed to be non-normally distributed. Treatments were categorized as curative, endoarterial, or systemic, and analyzed across up to six treatment courses. The data did not report missing values; therefore, a complete case analysis was performed. Patients were censored at liver transplantation (LT), death, or last follow-up. Missing death dates were sourced from hometown registries. Cancers 2024,16, 3549 4 of 16 Survival probabilities were calculated using the Kaplan–Meier method, with the logrank test comparing different treatments. The primary survival analysis was by type of treatment group and the secondary survival analysis was by presence or absence of nTS. Univariable and multivariable Cox proportional hazards regression analyses were used to identify factors associated with patients’ survival. Child–Pugh and MELD scores were not included to avoid collinearity. Baseline variables collected included age, sex, performance status, etiology, bilirubin, albumin, International Normalized Ratio, creatinine, ascites, encephalopathy, number of HCC nodules (one, two, three, or multiple), presence of portal vein thrombosis, and transcriptomic signature [ 15 ]. Five competing-risk Cox proportional hazards regression models were developed to account for the importance of different collinear factors. All statistical tests were performed with a two-sided significance threshold set at p< 0.05. miRNA targets were predicted by means of three tools as miRWalk (http://mirwalk. umm.uni-heidelberg.de/) [ 38 ], miRDB (http://www.mirdb.org/) [ 39 ], and miRabel (http: //bioinfo.univ-rouen.fr/mirabel/) [40]. PASW Statistics (ver. 28; IBM Corporation, Armonk, NY, USA) was used for statistical analysis. 3. Results 3.1. Clinical Results Three hundred and twenty-eight patients were enrolled in this prospective study. Median observation time was 31 months (mean ± SD: 40.5 ± 35.2 months). Patients enrolled in this study were mostly males (n= 259, 79.0%). The mean age at diagnosis was 65.1 ± 11.2 (median 67) years. Viral etiology was prevalent (64.9%; of these, 53.9% HCV-positive and 11.0% HBV-positive). Alcohol abuse was the only etiologic factor in 16.5% of subjects and was associated with HCV in 9.1% of patients. NASH was present in 18.3% of cases. Liver function was preserved in most patients (69.2% in Child–Pugh Class A, MELD score of 10.5 ± 3.8, median 9.0). nTS+ patients more often had multifocal presentation (nTS+ vs. nTS − : 39.4% vs. 18.3%; p< 0.001), were more often Edmondson– Steiner grade 3 or 4 (74.1% vs. 57.3%, p= 0.028), and had higher AFP levels (3.896 ± 13.265 vs. 954 ±6.209 ng/mL ,p= 0.012). Composition of derivation and validation cohorts was similar (Supplementary Table S1). 3.2. Therapeutic Management and Outcome At diagnosis, 15.9% (52 patients) received only supportive care, with no significant difference between nTS − and nTS+ groups. nTS+ patients were more likely to need systemic therapy from the onset (26.7% vs. 11.9%, p= 0.0012). A total of 276 patients (84.1%) received at least one HCC treatment, with 484 treatments administered overall (Supplementary Table S2). The median intervals between the first, second, and third treatments were 30, 33, and 22 weeks, respectively. Notably, fewer nTS+ patients proceeded beyond the second treatment. The average number of treatments was similar between nTS − and nTS+ groups. As initial treatments, resection and RFA were common, but nTS+ patients were significantly less likely to be suitable for RFA (p= 0.003). TACE was also more frequent in nTS − patients as a first treatment (p= 0.01). In subsequent treatments, the use of curative methods decreased, while endoarterial treatments remained high. Liver transplants were more common in later treatments but were still rare overall (n= 42, 8.6%). Among these, only 13.6% were nTS+ patients, often ineligible for transplant due to advanced disease or rapid progression. Cancers 2024,16, 3549 5 of 16 Three months post-treatment, around 36% of patients achieved an objective response, significantly more among nTS − compared to nTS+ patients (p= 0.0042). The cure rate was low, around 20%, with no significant difference between nTS − and nTS+ groups. These rates declined to 18.0% for responses and 8.5% for cures after the second treatment. Few patients progressed to a third treatment, with both response and cure rates diminishing further. Supplementary Figure S1 displays outcomes of sequential treatments, showing proportions of patients achieving cure receiving non-curative treatments like TACE, systemic therapy, or BSC. Figure S1B,C illustrate the outcomes for nTS − and nTS+ patients, emphasizing the significant differences in cure rates and mortality between the groups (p< 0.001). 3.3. Survival Analysis 3.3.1. Kaplan–Meier Analysis By this study’s end, 79.9% of participants had died, with a median survival of 31 months (Supplementary Figure S2A). Kaplan–Meier analysis revealed that nTS+ patients had a significantly shorter survival of 13 months, compared to 41 months for nTS − patients (Supplementary Figure S2B, p< 0.0001). This trend was consistent with that of the validation cohort (Supplementary Figure S2C,D). Twenty-six patients (7.9%) died after an average of 16 months (median 12 months) before they could receive a second treatment. The mortality rate following the first treatment was significantly higher in the nTS+ group compared to the nTS − group (33.7% vs. 16.5%; p= 0.004). Survival was significantly better for patients receiving any treatment, with a median survival of 44 months, compared to those receiving only BSC or systemic therapy, with median survivals of 7 and 11 months, respectively. Patients undergoing multiple consecutive treatments had a median survival of 57 months (p< 0.0001, log-rank test) (Figure 1A). Survival disparities were pronounced between nTS − and nTS+ patients across all treatment categories, with nTS+ patients experiencing poorer outcomes. Median survivals were 9 vs. 6 months for BSC, 19 vs. 9 months for systemic therapy, 47 vs. 20 months for at least one treatment, and 68 vs. 33 months for multiple treatments for nTS − vs. nTS+ patients (p< 0.0001, log-rank test) (Figure 1C,D,E,F), despite similar rates of multiple treatments between nTS+ and nTS − patients (24.4% vs. 29.8%; p= 0.202, Fisher’s exact test). The validation cohort confirmed these patterns, showing consistent treatment responses and survival rates for the whole cohort (Figure 1B) and in the treatment subgroups (Figure 1G,H,I,J). Overall survival across the entire cohort varied significantly depending on the type of dominant treatment administered, defined as the treatment with the highest potential therapeutic impact. LT yielded the longest median survival at 123 months, outpacing other treatments like RFA at 42 months, surgical resection at 52 months, TACE at 34 months, systemic therapy at 11 months, and BSC at 6 months (p< 0.001) (Figure 2A,B). For nTS − patients, survival benefits scaled up from supportive to more curative treatments. Their median survival times varied by treatment type: 123 months for LT, 76 months for resection, 40 months for TACE, 44 months for RFA, 17 months for systemic therapy, and 9 months for BSC. Conversely, nTS+ patients experienced limited benefits across the treatment spectrum, with liver transplant outcomes being less favorable (median survival of 34 months) compared to resection (33 months), RFA (52 months), TACE (29 months), systemic therapy (9 months), and BSC (5 months). A direct comparison between each type of treatment is shown in Figure 2C–H for the derivation cohort and in Figure 2I–N for the validation cohort. Cancers 2024,16, 3549 6 of 16 Figure 1. Outcome of treatments (BSC, systemic therapy, at least one treatment, and multiple treatments) according to presence or absence of the transcriptomic signature (nTS) in HCC. The outcomes for the derivation and validation cohorts as a whole are depicted in (A,B), respectively. Survival for best supportive care (BSC), systemic therapy, one therapeutic course, or multiple therapeutic courses, stratified for presence or absence of nTS, are depicted in (C–F) (derivation cohort) and (G–J) (validation cohort) (log-rank test). 3.3.2. Cox Regression Analysis We conducted a univariate Cox regression to assess the impact of factors like transcriptomic signature, performance status, liver transplant, BCLC stage, albumin, CRP, nodule count, and portal vein thrombosis on survival, all showing significant associations (Table 1). In multivariate analysis, we adjusted for collinearity, excluding CRP and creating five models that included either transcriptomic signature, portal vein thrombosis, or nodule number. The transcriptomic signature and the nodule number emerged as independent survival predictors. A liver transplant was a consistent positive predictor, and higher albumin levels also correlated with better survival in two models. Both the transcriptomic signature’s negative impact and the liver transplant’s positive impact were confirmed in the validation cohort. Cancers 2024,16, 3549 7 of 16 Figure 2. Outcome of treatments according to nTS status. Considering the outcome of all aggregate treatments performed in each patient, a clear-cut survival difference was present depending on the presence, in the totality of treatments performed, of a dominant treatment ((A), derivation cohort and (B), validation cohort). Stratification of the cohort by transcriptomic signature showed optimal results for LT and progressively worse results for less curative treatments in the nTS − patients ((C–H), derivation cohort and (I–N), validation cohort) (p< 0.001, log-rank test). Survival was much worse in nTS+ patients, although the difference among treatments was often significant (BCS: best supportive care; Systemic Tx: systemic treatment; TACE: transarterial chemoembolization; RFA: radiofrequency ablation; LT: liver transplantation). Table 1. The results from both univariate and multivariable Cox regression analyses, tested using the Wald method, highlight the significant impact of various factors on patient survival. The neoangiogenic transcriptomic signature (nTS) showed collinearity with several baseline factors, such as multifocal tumors, portal vein thrombosis (PVT), Edmondson–Steiner grading, AFP levels, and C-reactive protein (CRP) levels. To mitigate this, five distinct multivariable models were crafted: Model 1 for nTS, Model 2 for tumor multifocality, Model 3 for PVT, Model 4 for Edmondson–Steiner grading, and Model 5 for AFP levels. CRP was omitted from these models due to its collinearity with all variables but the Child–Pugh score, ensuring the accuracy of the prognostic evaluations for these factors on patient outcomes. Univariable Multivariable HR 95% CI pHR 95% CI p Model 1 Transcriptomic signature 2.660 1.873–3.790 <0.001 2.444 1.491–4.007 <0.001 Cancers 2024,16, 3549 8 of 16 Table 1. Cont. Univariable Multivariable HR 95% CI pHR 95% CI p Sex 1.136 0.831–1.553 0.425 Age at diagnosis 1.014 1.002–1.026 0.027 1.009 0.985–1.034 0.466 Performance status 2.010 1.086–3.718 0.026 1.140 0.580–1.816 0.703 LT vs. others 0.180 0.104–0.310 <0.001 0.130 0.055–0.303 <0.001 Surgical resection vs. others 0.448 0.280–1.020 0.001 0.588 0.337–1.025 0.061 Etiology (viral vs. nonviral) 1.248 0.963–1.618 0.094 Ascites 1.317 0.987–1.75 0.061 Encephalopathy 0.779 0.249–2.440 0.668 BCLC stage 1.482 1.140–1.926 0.003 1.215 0.824–1.972 0.275 Bilirubin 0.990 0.935–1.059 0.872 Albumin 0.787 0.626–0.989 0.040 0.690 0.472–1.032 0.071 Creatinine 1.475 0.938–2.320 0.093 CRP 1.094 1.031–1.161 0.003 INR 0.903 0.547–1.490 0.689 Multiple nodules at presentation 1.305 1.160–1.469 <0.001 PVT 1.676 1.185–2.371 0.004 Model 2 Multiple nodules at presentation 1.305 1.160–1.469 <0.001 1.346 1.136–1.565 <0.001 Performance status 2.010 1.086–3.718 0.026 0.974 0.535–2.081 0.877 Age at diagnosis 1.014 1.002–1.026 0.027 1.055 0.993–1.042 0.580 LT vs. others 0.180 0.104–0.310 <0.001 0.123 0.053–0.286 <0.001 Surgical resection vs. others 2.230 1.04–3.566 0.001 0.776 0.446–1.333 0.354 BCLC stage 1.482 1.140–1.926 0.003 1.094 0.796–1.504 0.580 Albumin 0.787 0.626–0.989 0.040 0.486 0.340–0.695 <0.001 Model 3 Portal vein thrombosis 1.676 1.185–2.371 0.004 1.927 0.879–4.223 0.102 Performance status 2.010 1.086–3.718 0.026 0.867 0.411–1.830 0.709 Age at diagnosis 1.014 1.002–1.026 0.027 1.055 0.993–1.042 0.580 LT vs. others 0.180 0.104–0.310 <0.001 0.089 0.037–0.212 <0.001 Surgical resection vs. others 2.230 1.04–3.566 0.001 1.344 0.780–2.31 0.287 BCLC stage 1.482 1.140–1.926 0.003 1.095 0.801–1.496 0.569 Albumin 0.787 0.626–0.989 0.040 0.538 0.378–0.765 <0.001 Model 4 Edmondson–Steiner grading 1.431 1.232–1.662 <0.001 1.555 0.178–2.052 0.002 Performance status 2.010 1.086–3.718 0.026 1.469 0.760–2.841 0.253 Age at diagnosis 1.014 1.002–1.026 0.027 1.006 0.983–1.030 0.614 LT vs. others 0.180 0.104–0.310 <0.001 0.141 0.064–0.310 <0.001 Surgical resection vs. others 2.230 1.04–3.566 0.001 1.380 0.803–2.369 0.244 BCLC stage 1.482 1.140–1.926 0.003 1.301 0.978–1.729 0.070 Albumin 0.787 0.626–0.989 0.040 0.477 0.334–0.680 <0.001 Cancers 2024,16, 3549 9 of 16 Table 1. Cont. Univariable Multivariable HR 95% CI pHR 95% CI p Model 5 AFP levels (median) 1.784 1.328–2.397 <0.001 1.440 0.930–2.230 0.102 Performance status 2.010 1.086–3.718 0.026 1.353 0.669–2.738 0.400 Age at diagnosis 1.014 1.002–1.026 0.027 1.003 0.979–1.028 0.812 LT vs. others 0.180 0.104–0.310 <0.001 0.123 0.053–0.286 <0.001 Surgical resection vs. others 2.230 1.04–3.566 0.001 1.344 0.780–2.31 0.287 BCLC stage 1.482 1.140–1.926 0.003 1.095 0.801–1.496 0.569 Albumin 0.787 0.626–0.989 0.040 0.538 0.378–0.765 <0.001 3.4. Effect of Repeat Treatments on Transcriptomic Signature and microRNA In a cohort of 105 TACE patients, 86 underwent the procedure twice, and 43 three times; for RFA, 26 out of 59 required a second treatment. Biopsies to assess transcriptomic shifts were performed on 80 TACE patients before their second treatment and on all patients before their second RFA treatment. After the first TACE, 21 patients (26.2%) shifted from nTS − to nTS+ status. This shift increased to 58.1% (25 of 43) after the second TACE, compared to only 7.7% (2 of 26) in RFA patients, showing a statistically significant difference between the treatments (p= 0.039). This transition in TACE patients significantly affected survival, with a median survival of 25 months for those who transitioned to nTS+ status, as opposed to 35 months for those who did not (p= 0.030). This became apparent 18 months post-treatment (Figure 3). Cancers 2024, 16, x FOR PEER REVIEW 9 of 16 Surgical resection vs. others 2.230 1.04–3.566 0.001 1.380 0.803–2.369 0.244 BCLC stage 1.482 1.140–1.926 0.003 1.301 0.978–1.729 0.070 Albumin 0.787 0.626–0.989 0.040 0.477 0.334–0.680 <0.001 Model 5 AFP levels (median) 1.784 1.328–2.397 <0.001 1.440 0.930–2.230 0.102 Performance status 2.010 1.086–3.718 0.026 1.353 0.669–2.738 0.400 Age at diagnosis 1.014 1.002–1.026 0.027 1.003 0.979–1.028 0.812 LT vs. others 0.180 0.104–0.310 <0.001 0.123 0.053–0.286 <0.001 Surgical resection vs. others 2.230 1.04–3.566 0.001 1.344 0.780–2.31 0.287 BCLC stage 1.482 1.140–1.926 0.003 1.095 0.801–1.496 0.569 Albumin 0.787 0.626–0.989 0.040 0.538 0.378–0.765 <0.001 3.4. Effect of Repeat Treatments on Transcriptomic Signature and microRNA In a cohort of 105 TACE patients, 86 underwent the procedure twice, and 43 three times; for RFA, 26 out of 59 required a second treatment. Biopsies to assess transcriptomic shifts were performed on 80 TACE patients before their second treatment and on all patients before their second RFA treatment. After the first TACE, 21 patients (26.2%) shifted from nTS− to nTS+ status. This shift increased to 58.1% (25 of 43) after the second TACE, compared to only 7.7% (2 of 26) in RFA patients, showing a statistically significant difference between the treatments (p = 0.039). This transition in TACE patients significantly affected survival, with a median survival of 25 months for those who transitioned to nTS+ status, as opposed to 35 months for those who did not (p = 0.030). This became apparent 18 months post-treatment (Figure 3). Figure 3. Kaplan–Meier analysis of the survival of HCC patients who received TACE as their first treatment. Repeat biopsies were obtained from 80 of the 86 patients who underwent a second TACE, for comparison with baseline. A significant worsening in median survival was observed with the transition from nTS− to nTS+ status (p = 0.030, log-rank test). The transition involved changes in tumor microRNA expression levels related to angiogenesis, proliferation, cell cycle, and hypoxia. Initial microRNA levels were similar between transarterial chemoembolization (TACE) and radiofrequency ablation (RFA) patients, but significant changes occurred after treatment, especially post-TACE. Most alterations were seen after the second TACE biopsy, while minimal changes followed RFA treatment. Notably, TACE-related changes were mostly unfavorable, whereas RFA-induced alterations, such as increased levels of miR-145-5p or miR-30a-5p, were neutral or beneficial (Figure 4). Figure 3. Kaplan–Meier analysis of the survival of HCC patients who received TACE as their first treatment. Repeat biopsies were obtained from 80 of the 86 patients who underwent a second TACE, for comparison with baseline. A significant worsening in median survival was observed with the transition from nTS−to nTS+ status (p= 0.030, log-rank test). The transition involved changes in tumor microRNA expression levels related to angiogenesis, proliferation, cell cycle, and hypoxia. Initial microRNA levels were similar between transarterial chemoembolization (TACE) and radiofrequency ablation (RFA) patients, but significant changes occurred after treatment, especially post-TACE. Most alterations were seen after the second TACE biopsy, while minimal changes followed RFA Cancers 2024,16, 3549 16 of 16 46. Lasagni, S.; Leonardi, F.; Pivetti, A.; Di Marco, L.; Ravaioli, F.; Serenari, M.; Gitto, S.; Critelli, R.M.; Milosa, F.; Romanzi, A.; et al. Endothelial angiopoietin-2 overexpression in explanted livers identifies subjects at higher risk of recurrence of hepatocellular carcinoma after liver transplantation. Front. Oncol. 2022,12, 960808. [CrossRef] 47. Di Marco, L.; Pivetti, A.; Foschi, F.G.; D’amico, R.; Schepis, F.; Caporali, C.; Casari, F.; Lasagni, S.; Critelli, R.M.; Milosa, F.; et al. Feasibility, safety, and outcome of second-line nivolumab/bevacizumab in liver transplant patients with recurrent hepatocellular carcinoma. Liver Transplant. 2023,29, 559–563. [CrossRef] 48. de Conti, A.; Ortega, J.F.; Tryndyak, V.; Dreval, K.; Moreno, F.S.; Rusyn, I.; Beland, F.A.; Pogribny, I.P. MicroRNA deregulation in nonalcoholic steatohepatitis-associated liver carcinogenesis. Oncotarget 2017,8, 88517–88528. [CrossRef] 49. Milosa, F.; Critelli, R.M.; Lasagni, S.; Pivetti, A.; Di Marco, L.; Romagnoli, D.; Carulli, L.; Dituri, F.; Mancarella, S.; Giannelli, G.; et al. Prognostic significance of hypoxic and metabolic gene profiling in hepatocellular carcinoma. Liver Cancer Int. 2021,2, 15–26. [CrossRef] 50. Lencioni, R.; Kudo, M.; Erinjeri, J.; Qin, S.; Ren, Z.; Chan, S.; Arai, Y.; Heo, J.; Ahn, M.; Escobar, J.; et al. EMERALD-1: A phase 3, randomized, placebo-controlled study of transarterial chemoembolization combined with durvalumab with or without bevacizumab in participants with unresectable hepatocellular carcinoma eligible for embolization. JCO 2024,42, LBA432. [CrossRef] 51. Wang, B.; Xu, H.; Gao, Z.Q.; Ning, H.F.; Sun, Y.Q.; Cao, G.W. Increased expression of vascular endothelial growth factor in hepatocellular carcinoma after transcatheter arterial chemoembolization. Acta Radiol. 2008,49, 523–529. [CrossRef] [PubMed] 52. Xiao, E.H.; Guo, D.; Bian, D.J. Effect of preoperative transcatheter arterial chemoembolization on angiogenesis of hepatocellular carcinoma cells. World J. Gastroenterol. 2009,15, 4582–4586. [CrossRef] [PubMed] 53. Brown, J. SR 4233 (Tirapazamine): A new anticancer drug exploiting hypoxia in solid tumours. Br. J. Cancer 1993,67, 1163–1170. [CrossRef] [PubMed] 54. Lin, W.H.; Yeh, S.H.; Yeh, K.H.; Chen, K.W.; Cheng, Y.W.; Su, T.H.; Jao, P.; Ni, L.C.; Chen, P.J.; Chen, D.S. Hypoxia-activated cytotoxic agent tirapazamine enhances hepatic artery ligation-induced killing of liver tumor in HBx transgenic mice. Proc. Natl. Acad. Sci. USA 2016,113, 11937–11942. [CrossRef] [PubMed] 55. Sonoda, A.; Nitta, N.; Ohta, S.; Nitta-Seko, A.; Nagatani, Y.; Takahashi, M.; Murata, K. Enhanced Antitumor Effect of Tirapazamine Delivered Intraperitoneally to VX2 Liver Tumor-Bearing Rabbits Subjected to Transarterial Hepatic Embolization. Cardiovasc. Interv. Radiol. 2011,34, 1272–1277. [CrossRef] 56. Liu, C.-H.; Peng, C.-M.; Hwang, J.-I.; Liang, P.-C.; Chen, P.-J.; Abi-Jaoudeh, N.; Giiang, L.-H.; Tyan, Y.-S. Phase I Dose-Escalation Study of Tirapazamine Chemoembolization for Unresectable Earlyand Intermediate-Stage Hepatocellular Carcinoma. J. Vasc. Interv. Radiol. 2022,33, 926–933.e1. [CrossRef] 57. Abi-Jaoudeh, N.; Dayyani, F.; Chen, P.J.; Fernando, D.; Fidelman, N.; Javan, H.; Liang, P.-C.; Hwang, J.-I.; Imagawa, D.K. Phase I Trial on Arterial Embolization with Hypoxia Activated Tirapazamine for Unresectable Hepatocellular Carcinoma. J. Hepatocell. Carcinoma 2021,8, 421–434. [CrossRef] Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.