Targeting the mTOR pathway in hepatocellular carcinoma: Current state and future trends

Summary

Mechanistic target of rapamycin (mTOR) regulates cell growth, metabolism and aging in response to nutrients, cellular energy stage and growth factors. mTOR is frequently up-regulated in cancer including hepatocellular carcinoma (HCC) and is associated with bad prognosis, poorly differentiated tumors, and earlier recurrence. Blocking mTOR with rapamycin and first generation mTOR inhibitors, called rapalogs, has shown promising reduction of HCC tumors growth in preclinical models. Currently, rapamycin/rapalogs are used in several clinical trials for the treatment of advanced HCC, and as adjuvant therapy in HCC patients after liver transplantation and TACE. A second generation of mTOR pathway inhibitors has been developed recently, and is being tested in various clinical trials of solid cancers and has been used in preclinical HCC models. The results of series of clinical trials using mTOR inhibitors in HCC treatment will emerge in the near future.

Introduction

Target of rapamycin (TOR) is an evolutionary well conserved serine/threonine protein kinase that belongs to the phosphoinositide 3-kinase (PI3K)-related kinase family. Mechanistic TOR (mTOR; originally called mammalian TOR) has a broad range of action and is involved in regulation of cell growth, aging and metabolism¹. mTOR can be divided into two structurally and functionally distinct complexes named mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2)¹. mTORC1 is composed of mTOR, mLST8, DEPTOR, RAPTOR and PRAS40. mTORC2 consists of mTOR, mLST8, DEPTOR, PROTOR, RICTOR and mSIN1¹. mTORC1 is a nutrient and energy sensor at both cellular and whole-body levels². When nutrients are available, mTORC1 is activated and stimulates anabolic processes such as protein synthesis, lipogenesis, and energy metabolism, whereas autophagy and lysosome biogenesis is inhibited¹ (for more details see Figure 1). mTORC1 is activated by a myriad of inputs such as growth factors, energy status, proinflammatory cytokines, oxygen levels, amino acids, and the canonical Wnt pathway¹ (Figure 1). Growth factors, e.g. insulin and insulin-like growth factor 1 (IGF1), exert their action on mTORC1 through receptor tyrosine kinases (RTK) and the well-characterized PI3K-AKT and Ras-Raf-Mek-Erk signaling pathways. These pathways activate mTORC1 by phosphorylating and thereby inhibiting the tumor suppressor TSC1-TSC2 (tuberous sclerosis 1 and 2) complex. The TSC1-TSC2 complex is a key regulator of mTORC1 and functions as a GTPase-activating protein (GAP) that negatively regulates Rheb by converting it into its inactive GDP-bound state^{3, 4}. In contrast, down-regulation of mTORC1, is accomplished via activation of the TSC1-TSC2 complex by AMPK, LKB1 and REDD1 in situations of low energy (high AMP), low oxygen levels⁵ and DNA damage⁶.

Figure 1.

Schematic overview of the mTOR signaling pathway with the most important factors and their action.

Much less is known about the later discovered mTORC2 signaling pathway. mTORC2 is insensitive to nutrients but does respond to growth factors such as insulin in association with ribosomes⁷. Besides its initial described role in actin cytoskeleton organization, mTORC2 also activates cell metabolism, survival, and growth. TORC2-ribosome interaction is a likely conserved mechanism of TORC2 activation that is physiologically relevant in both normal and cancer cells.

Involvement of mTOR pathway in hepatocellular carcinoma (HCC)

Given its importance in cell growth and metabolism it is not surprising that mTOR plays a pivotal role in HCC. mTORC1 and mTORC2 pathways, including pRPS6, p-AKT, IGF-1R and RICTOR are up-regulated in 40-50% of HCCs^8–10. A similar upregulation is observed in other common cancer types such as breast, colon and lung carcinomas¹¹. Moreover an up-regulation is frequently observed in cholangiocarcinoma, the second most common primary cancer of the liver¹². Activation of the mTOR pathway in HCC is associated with less differentiated tumors, bad prognosis, and earlier recurrence independently of the underlying etiology of liver cancer^{9, 13, 14}. Furthermore, it is associated with deregulation of EGF, IGF and PTEN pathways⁹ and, as expected, with increased lipogenesis in the tumor¹⁵. Surprisingly, alterations in copy number or somatic mutations of PTEN, PIK3CA, and PIK3B were not identified as major mechanisms of mTOR pathway deregulation in HCC by PCR⁹. In accordance, more recent studies using next-generation sequencing technique revealed a low frequency of mutations in the mTOR pathway including mTOR, PIK3CA and PTEN among others^16–18. The most frequently mutated gene, found in one study in 9.6% of HCC was RPS6KA3, a serine/threonine kinase involved in regulating PI3K/RAS signaling¹⁶. Therefore, mutations in the mTOR pathway is a rare event in HCC and activation of the mTOR pathway appears to result largely from ligand dependent receptor activation.

Genomic studies in the past have identified multiple molecular classifications of HCC and demonstrated deregulated signaling pathways unique to subgroups of patients^19–24. These studies indicated that the mTOR pathway and its upstream pathways PI3K and AKT occupy a central position in the network of deregulated signaling pathways in HCC. With the specific aim to identify driver genes associated with HCC prognosis, we have used an integrative approach combining data obtained from somatic copy number analysis and transcriptomics²⁵. Fifty driver genes were recognized and were linked to the mTOR, AMPK or EGFR pathways. In the molecular HCC classification by Boyault et al., 6 robust subgroups (G1-G6) were identified and the G1/G2 subgroup showed AKT activation with overexpression of IGF2, IGF1R and GSK3β as well as PIK3CA, and AXIN1 mutations¹⁹. The G1/G2 patient subgroup was further confirmed in a large meta-analysis using integrative transcriptomics of 9 HCC data sets including a total of 603 patients²⁶. This analysis assigned the patients into three subclasses (S1-S3), and the G1/G2 subgroup was enriched in the subclass S2, characterized again by activation of the upstream regulator of mTOR, AKT, in combination with MYC.

Taken together, activation of mTOR plays a central role in HCC and blocking this pathway is an attractive strategy for HCC treatment. The main goal of this review is to offer the rationale for the use of mTOR inhibitors in HCC and provide an overview of the current and prospective clinical trials with mTOR inhibitors in HCC.

Rapamycin and first generation mTOR inhibitors

mTOR is targeted by rapamycin, a natural compound discovered from the bacterium Streptomyces hygroscopius more than 30 years ago. The two mTOR-containing complexes have different sensitivities to rapamycin. mTORC1 is inhibited by a complex formed by rapamycin and FKBP12 protein²⁷. In contrast, mTORC2 is generally resistant to rapamycin, however, in certain cell types, mTORC2 may show sensitivity after prolonged rapamycin treatment²⁸. Rapamycin (sirolimus) was first approved as an immunosuppressant for the prevention of graft rejection in kidney transplant recipients more than a decade ago². A few years later rapamycin obtained approval for its use as an anti-restenosis agent following balloon angioplasty in coronary arterial stents. The early success of rapamycin has encouraged the development of derivative compounds with improved bioavailability, called rapalogs: everolimus (RAD001), temsirolimus (CCI-779), and deforolimus (AP23573). Due to the important role of mTOR in cell growth and metabolism the primary interest shifted to anti-cancer therapy, and in 2007 temsirolimus (CCI779) was approved for the treatment of renal cell carcinoma and shortly thereafter for mantle cell lymphoma. Meanwhile, everolimus (RAD001), has received approval for treatment of pancreatic neuroendocrine tumours, subependymal giant cell astrocytoma, renal cell carcinoma and HER2-negative breast cancer in combination with Exemestane.

In general, first generation mTOR inhibitors are well tolerated. The major toxicities include, stomatitis, headache, diarrhea, vomiting and thrombocytopenia. Due to their role in metabolism they can cause hyperglycemia, hyperlipidemia and hypophosphatemia. As for every immunosuppressive drug, the risk for infections is increased. Furthermore, reactivation of HBV is a serious complication and has been described in renal cell carcinoma patients under everolimus treatment^{29, 30}. EASL Clinical practice Guidelines recommend in which situation patients undergoing immunosuppressive therapy should receive prophylactic treatment against HBV-reactivation with a nucleoside analogue³¹.

Currently, neither rapamycin nor rapalogs have gained approval for HCC treatment. However, several clinical trials are ongoing or have recently been completed using rapamycin/rapalogs for the treatment of advanced HCC and as adjuvant therapy after transarterial chemoembolization (TACE) or after liver transplantation of HCC patients. Furthermore, several studies suggest that mTOR inhibition may even prevent HCC development in patients at risk. The application of mTOR inhibitors in these different settings will be discussed in more detail.

1. Prevention of liver cancer by mTOR inhibition

Several studies indicate that mTOR activation is involved in the initiation of liver cancer and plays a role in the malignant transition of hepatocytes to HCC. A gradual activation of the AKT/mTOR pathway was progressively induced from non-tumorous liver tissue toward the HCC thus supporting this concept¹⁵. Interestingly, increased mTOR activity conferred a preneoplastic phenotype to HepaRG, considered a terminally differentiated hepatic cell line³². Also, treatment with mTOR inhibitor everolimus prevented proliferation of hepatocytes suffering DNA damage in the fumarylacetoacetate hydrolase-deficient mouse model of chronic liver injury and HCC development³³. In transgenic mice, mTOR activation by itself was shown to be sufficient for HCC development. Two independent studies analyzed liver-specific knockout of TSC1 and the resulting chronic activation of mTOR^{34, 35}. Both studies demonstrated the development of liver tumors, however, only one was associated with inflammation³⁴. Likewise constant activation of mTOR in PTEN-deficient mice induced steatohepatitis and development of liver tumors³⁶. mTOR inhibitors may also prevent HCC indirectly by reducing liver fibrosis, a risk factor for the development of HCC. Studies in rats have demonstrated that sirolimus and everolimus attenuated progression of fibrosis, in contrast to cyclosporine A and tacrolimus, two other immunosuppressive drugs^{37, 38}.

Randomized prospective clinical trials have not been conducted to address the question if mTOR inhibitors prevent HCC. The evidence suggesting a potential HCC prevention is solely derived from epidemiological studies with metformin, a widely used anti-diabetic drug that reduces mTOR activity but importantly also ameliorates hyperinsulinemia, which is a risk factor for HCC³⁹. Metformin activates AMPK, which in turn suppresses the mTORC1 pathway (Fig. 2)⁴⁰. A recent study showed that metformin controlled gene expression at the level of mRNA translation to an extent comparable to that of canonical mTOR inhibitors and down regulated mRNAs, which encode for proliferation and tumor-promoting proteins via the mTORC1/4E-BP pathway⁴¹. In addition, it was shown that metformin inhibited mTORC1 signaling independently of AMPK by suppressing Rag GTPase or activating REDD1^{42, 43}. A meta-analysis, including 5 clinical studies with more than 100 000 type 2 diabetes patients has been performed lately⁴⁴. This study showed an overall estimated 62% reduction in the risk of liver cancer if metformin was used as an anti-diabetic treatment instead of non-metformin treatment (e.g. sulfonylurea or insulin). Although there was considerable heterogeneity between the studies, the results are encouraging. Preventive clinical trials with metformin are already underway for different cancer types such as breast cancer, colon cancer and esophagus cancer, however, for HCC they are lacking. Information of the potential preventive mechanism(s) of metformin on liver cancer is also limited. Recently, it was shown in vitro and in vivo, that metformin inhibited HCC cell growth through AMPK and LKB1^{45, 46}. Further elucidation of this mechanism(s) and defining criteria that identify individuals which are likely to benefit are therefore needed. Nevertheless, it has to be kept in mind that a few studies have shown that metformin can have tumor-promoting effects. Metformin enhanced growth of BRAF-mutant melanoma cells and ER-alpha negative breast cancer cell lines in vivo^{47, 48}. Likewise, migration and invasion abilities of human pulmonary adenocarcinoma A549 cell lines increased under metformin treatment in vitro⁴⁹.

Figure 2.

Schematic overview of the mTOR signaling pathway with the target position of drugs.

Prevention of HCC by metformin and mTOR inhibitors may also involve autophagy, an important homeostatic cellular recycling mechanism, which has a dual role in carcinogenesis. Autophagy is important in limiting DNA damage caused by the accumulation of reactive oxygen species and damaged organelles through removal of dysfunctional proteins and organelles⁵⁰. Especially in the context of HCC development, which arises in 80% of the cases in the background of a chronic liver inflammation with constant DNA damage, induction of autophagy by mTOR inhibitors may prevent HCC development and recurrence. In contrast, increased autophagy may promote tumor growth, especially once tumor is established, as autophagy is also a pro-survival mechanism to cellular stress and further enhances chemoresistance^{50, 51}.

2. Adjuvant therapy with mTOR inhibitors after liver transplantation or TACE

For patients with non-resectable HCC and early stage disease, liver transplantation is recognized as the treatment of choice, with disease-free survival of 60-80% at 5 years⁵². Immunosuppressants, such as cyclosporine A, tacrolimus and rapamycin are crucial for the prevention of graft rejection after transplantation. However, cyclosporine A and tacrolimus are known to induce tumor growth in preclinical studies^{53, 54}. On the contrary, rapamycin has additional anticancer function. Accordingly, a meta-analysis of 5 studies with a total of almost 3000 patients indicated that survival is significantly prolonged after liver transplantation of HCC patients if sirolimus is administered instead of non-sirolimus immunosuppressives⁵⁵. Although these results indicate that sirolimus should be used as treatment of choice after liver transplantation of HCC patients, it has to be noted that none of the clinical studies conducted so far were randomized. Additional information regarding the efficacy of sirolimus to improve survival and prevent recurrence should be provided by three ongoing randomized controlled studies (Table 1). The largest trial is a multicenter, phase III study with over 500 patients (SiLVER) comparing sirolimus-based versus sirolimus-free immunosuppression in patients undergoing liver transplantation for HCC. However, results are not expected before 2014.

Table 1.

Summary of completed and ongoing clinical trials with first generation mTOR inhibitors in HCC

			Drug	Patient number	HCC Stage	Phase	Child- Pugh Score	Study design	Results (or ID in Clinical Trials website)
Adjuvant therapy after liver transplantation			Sirolimus vs mTOR inhibitor free	510	Milan criteria and extended	III	NA	Randomized	Active, not recruiting (NCT00355862, SiLVER)
			Sirolimus vs FK506	220	Exceeding Milan criteria	III	NA	Randomized	Recruiting (NCT00554125)
			Sirolimus vs mTOR inhibitor free	86	Exceeding Milan criteria	II	NA	Randomized	Recruiting (NCT01374750)
Adjuvant therapy after TACE			TACE (Doxorubicin) +/− everolimus	98	Intermediate	I/II	A, B (≤7)	Randomized	Recruiting (NCT01009801)
			TACE +/− Everolimus	80	Intermediate	II	A, early B	Randomized	Recruiting (NCT01379521, TRACER)
Treatment of advanced HCC	Single drug	First line therapy	Sirolimus	21	I to IV (TNM)	pilot	A, B, C	Non-randomized	Completed63. 1 PR Dose determined by plasma level. Median dose was 1mg/d;
			Sirolimus	18	BCLC B, C, D	pilot	A, B, C	Non-randomized	Completed62. No objective response Twice daily. Dose determined by plasma level. Median dose was not reported.
			Sirolimus	25	BCLC B or C	II	A, B	Non-randomized	Completed64. 1 CR and 1 PR, 30mg/week;
			Temsirolimus	50	advanced	II	B	Non-randomized	Terminated: toxic events
		First and second line	Everolimus	28	advanced	I/II	A, B	Non-randomized	Completed60. 1 PR MTD = 10mg/day;
			Everolimus	39	III (AJCC)	I/II	A, B	Randomized	Completed61. 1 PR MTD = 7.5mg/day or 70mg/week but DLT not reach for weekly schedule;
		Second line therapy	Everolimus	546	advanced	III	A	Randomized	Completed, not yet published, failed to demonstrate efficacy compared to placebo (NCT01035229, EVOLVE)
			Temsirolimus	50	advanced	I/II	A	Non-randomized	Active, not recruiting (NCT01251458)
			Temsirolimus	25	advanced	II	A, B (≤9)	Non-randomized	Recruiting (NCT01567930)
	Drug combination	First and second line	Sorafenib and Temsirolimus	25	III, IV	I	A, B (≤7)	Non-randomized	Completed70. 2 PR; MTD = temsirolimus 10mg/week + sorafenib 200mg twice daily;
		First line therapy	Sorafenib +/− Everolimus	130	Advanced liver cancer	I/II	A	Non-randomized	Terminated: everolimus MTD too low
			Sorafenib +/− Everolimus	106	B,C BCLC	II	A, B (≤7)	Randomized	Active, not recruiting (NCT01005199)
			Sorafenib and Everolimus	30	III, IV and liver dysfunction	I	A, B (≤7)	Non-randomized	Active, not recruiting (NCT01013519)
			Sorafenib and Everolimus	28	AJCC II,III,IV	II	A, B (≤7)	Non-randomized	Recruiting (NCT01687673)
		Second line therapy	Everolimus+ Bevacizumab	24	advanced	I/II	A, B	Non-randomized	Completed. Not yet published (NCT00775073)
			Sirolimus + Bevacizumab	24	advanced	I	A, B	Non-randomized	Completed69. 1 CR and 2 PR; MTD = 4mg/day,
			Sorafenib and Everolimus	18	advanced	I/II	A, B (≤8)	Non-randomized	Recruiting (NCT01335074)
			Everolimus and Pasireotide	30	BCLC stage C	II	A	Non-randomized	Recruiting (NCT01488487)

Recruiting; the study is currently recruiting participants,

Active not recruiting: The study is ongoing but potential participants are not currently recruited

Completed: The study has ended normally, and participants are no longer being examined or treated

Terminated: The study has stopped recruiting participants early and will not start again

PR: partial response; CR: complete response

TACE is the treatment of choice for intermediate HCC patients⁵². Following TACE treatment the levels of vascular endothelial growth factor (VEGF) have been shown to increase. VEGF promotes angiogenesis and is associated with bad prognosis. Therefore, TACE has been combined in clinical trials with other drugs that inhibit angiogenesis such as sorafenib, brivanib or orantinib. Interestingly, everolimus has also been shown to reduce vessel formation in preclinical HCC models⁵⁶ and two ongoing clinical trials are evaluating the effect of TACE in combination with everolimus (Table 1).

3. mTOR inhibition in advanced HCC

Sorafenib is the only approved drug for HCC treatment^{57, 58}. However, the treatment effects are small, only selected patients are eligible for therapy, and side effects often limit applicability. Therefore, developing novel and effective therapies are urgently needed. Rapalogs have been shown to inhibit liver tumor growth in a large number of in vitro and in vivo pre-clinical studies⁵⁹ and have encouraged clinical trials in HCC patients (Table 1, information retrieved form www.clinicaltrials.gov). Two phase I/II dose-finding studies of 39 and 28 patients have been completed, using everolimus as a first and second line single agent and resulted in dose recommendations of 7.5mg/day and 10mg/day respectively^{60, 61}. In both studies, complete response was not observed, however, one HCC patient in each study showed a partial response, and stable disease was observed for a short time in 71.4% and 40% of the patients. Three earlier clinical studies with 18 to 25 patients using rapamycin as a first line single agent provided interesting results with partial responses and even one complete response; however, the differences in treatment schedule prevent any firm conclusions^62–64. Overall, these preliminary results using rapamycin/rapalogs are encouraging; however, further studies with more patients are needed. Besides, one study using temsiroliums as a first line agent had to be terminated due to toxic events. The recent outcome of a multicenter randomized, double blind, phase III study dashed the hope to use mTOR inhibitors as a second line therapy for advanced HCC patients (press release). This study investigated the effect of everolimus or placebo in 546 patients with Child-Pugh A cirrhosis, whose disease progressed after treatment with or who were intolerant to sorafenib. It will be interesting to see if two other phase I/II studies using temsirolimus as a second line single agent obtain similar negative results.

Because of resistance and compensatory activation of other signaling pathways the effect of rapalogs can be diminished. For example, after treatment with everolimus, an upregulation of MAPK was shown in tumor samples from breast cancer patients⁶⁵. As anticipated, combination of mTOR inhibition with a MAPK inhibitor (sorafenib) showed enhanced anti-tumoral effect in vitro and in vivo in cancer models including HCC^65–68. A combinatorial approach is currently performed in 9 clinical trials (Table 1). Rapalogs or rapamycin is complemented with multikinase inhibitor sorafenib, VEGF inhibitor bevacizumab or with pasireotide, a somatostatin analog which has been shown to inhibit tumor growth. Two phase I combination studies have been published recently^{69, 70}. A combination of temsirolimus with sorafenib at the maximal tolerated dose (MTD; temsirolimus 10mg weekly and sorafenib 200mg twice daily), showed a partial response in 8% of the patients and a stable disease in 60%⁷⁰. Although these results in 25 patients are promising, the median progression free survival was only 5.65 months, which is similar to outcomes from single agent sorafenib in the SHARP trial, though superior to outcomes observed in an Asia-Pacific clinical trial^{57, 58}. In addition, a phase I study with rapamycin and bevacizumab of 24 patients at MTD (rapamycin 4mg/day and bevacizumab 5 mg/kg every 14 days), reported a remarkable complete response in one patient that lasted 4.5 months, partial response in two patients, and stable disease in 14 patients⁶⁹. The ongoing clinical trials in phase I or II will reveal if a combinatorial approach improves efficacy in HCC treatment. The combination of drugs is certainly attractive, however, one drawback is that toxicities may increase, especially in cirrhotic patients. and prevent that efficacious dose can be applied. It is therefore important to obtain pharmacokinetic studies of both drugs (e.g. sorafenib and raplogs) with dose escalation of both drugs in cirrhotic patient in future clinical trials.

Second generation mTOR inhibitors

We are currently awaiting the results of several clinical trials using rapalogs in HCC treatment; however, the success of rapalogs in cancer therapy in general has not been as impressive as initially hoped. Several possible reasons may account for the limited action of rapalogs. First, mTORC1 inhibition abrogates the negative feedback loop, which in turn activates PI3K-AKT with MAPK and RAS signaling and therefore may actually increase growth of cancer cells⁶⁵. Second, blocking of mTORC1 primarily leads to inhibition of cell growth and not cell death. Third, mTOR inhibition by rapalogs mainly results in inhibition of S6K, however, the second key substrate, 4E-BP1, is only insufficiently blocked⁷¹. Finally, rapalogs do not inhibit mTORC2, which is often activated as part of the PI3K-mTORC2-AKT signaling axis². In order to overcome the shortcomings and resistance of rapalogs a second generation of mTOR inhibitors has been developed that function as ATP-competitive inhibitors of mTOR and have several advantages over rapalogs. Unlike rapalogs, which inhibit only mTORC1, the ATP analogues block the phosphorylation of all known downstream targets of mTORC1 and mTORC2. Furthermore, because of the similarity between the kinase domains of mTOR and PI3Ks, some of these new compounds additionally inhibit PI3K, leading to a broad inhibitory action with blocking of the feedback activation of PI3K-AKT signaling described before. Second generation mTOR inhibitors can therefore be divided into mTORC1/2 inhibitors and mTOR/PI3K inhibitors (Figure 2). In addition, a series of compounds have been developed that block upstream of the mTOR pathway such as AKT inhibitors and PI3K inhibitors.

1. Translational genomics and development of second generation mTOR inhibitors

Translational genomic analyses have been used to investigate the resistance to rapamycin. Jimenez et al.⁷² investigated rapamycin sensitivity or resistance in 13 HCC cell lines using drug-induced growth inhibition as the end point. The authors concluded that promoting or even maintaining effective drug sensitivity might be challenging, because of the molecular heterogeneity observed in the genomic profiles introduced in response to rapamycin, as well as determining a specific mechanism(s) of drug resistance. Furthermore, although the sensitivity to rapamycin was variable in all cell lines, the drug inhibited the phosphorylation of RPS6 and 4E-BP1, indicating that S6 and 4E-BP1 phosphorylation is not a useful marker for antiproliferative effect of mTOR inhibitors. Interestingly, in an attempt to identify rapamycin sensitive genes a later study determined that many of the genes which expression is altered by rapamycin are E-box containing and their regulation via mTOR was c-MYC independent⁷³. Since the mTOR pathway is constitutively activated in a majority of diffuse large B-cell lymphoma this is an interesting disease model to study resistance mechanisms to mTOR inhibitors and to identify drugs which may compliment the action of rapamycin⁷⁴. The authors compared the genomic signatures of 4 rapamycin sensitive and 4 resistant cell lines, and found that the central mechanism involved in the resistance to rapamycin was controlled by AKT. Using the genomic signatures to explore the Connectivity Map database to identify drugs which may reverse drug resistance, PI3K/AKT and HDAC inhibitors were the most likely candidates to synergize with an mTOR inhibitor. Resistance mechanisms to rapamycin and its derivates also include epigenome based reprogramming (e.g. miRs). Long-term rapamycin treatment results in an increase of the miR-17-92 cluster and inhibition of this change restored the drug sensitivity⁷⁵. Likewise, in HCC the miR-216a/217 cluster is frequently upregulated, resulted in activation of the PI3K/AKT pathway and importantly resistance to sorafenib⁷⁶.

2. Preclinical studies of second mTOR, AKT and PI3K inhibitors in HCC

Several preclinical studies with second generation mTOR, AKT, and PI3K inhibitors were performed in HCC cell lines^77–79, HCC xenograft mouse model^{77, 79}, DEN-mouse HCC model⁸⁰ and in a diabetic rat model of HCC⁸¹. These studies have demonstrated that the second generation inhibitors were able to control HCC proliferation better than everolimus or sirolimus. It has also been demonstrated that combination of a PI3K inhibitor (BKM120) with cisplatin⁷⁷ or a PI3K/mTOR dual inhibitor (BEZ235) with everolimus⁸⁰ could act synergistically with strong antitumor activity. The two drugs, everolimus and BEZ235 exerted tumor regression via inhibition of mTORC1 and mTORC2. This combined drug-effect was associated with an increase in autophagy independent of 4E-BP1. At low doses both drugs targeted mTORC1, however, inhibition of mTORC2 was enhanced by the drug combination. The cooperative drug-effect was further evident from the microarray analysis identifying a distinct set of genes, suggesting a phenotypic reversal similar to placebo-treated livers. Also, only the drug combination achieved significant inhibition of genes involved in cell cycle. These results have prompted a dose finding and a safety clinical trial of BEZ235 in combination with everolimus in patients with advanced solid tumors.

Moreover, promising result may be expected from the combination of second generation mTOR, AKT or PI3K inhibitors with other drugs. This may be particularly important in cases of sorafenib resistance which was recently demonstrated in a cancer stem cell subpopulation of HCC cells (i.e. label-retaining cancer cells), demonstrating sustained AKT and MAPK activation⁸². Also, in a study comparing the HCC cell line (Huh7) and sorafenib resistance Huh7 derivatives, the molecular alteration related to the acquired drug resistance included upregulation and activation of PI3K/AKT signaling⁸³. The sorafenib resistance was overcome by either silencing AKT using RNAi or targeting the pathway using the novel allosteric AKT inhibitor (MK-2206). In agreement, the PI3K/mTOR inhibitors (PI103 and PKI-587) also augment the effect of sorafenib^{84, 85}. Similarly, a recent study suggested that second-generation inhibitors (e.g. mTOR (AZD-8055), PI3K (BKM-120) or mTOR/PI3K (BEZ-235 and GDC-0980)) may be effective in sorafenib resistant HCC⁸⁶. It should be noted that this study unfortunately was performed in SK-HEP1 and its drug-resistant derivative (SK-Sora) which are derived from a HCC patient with ascites but are of endothelial origin. The combination of everolimus together with an AKT inhibitor (MK-2206), mTOR/PI3K inhibitor (BEZ235) or PI3K inhibitor (BKM120) showed improved anti-tumoral activity^{77, 78, 80}. Further, improved efficacy was also achieved in vitro and in vivo with novel ATP-competitive mTOR kinase inhibitors together with histone deacetylase inhibitors⁸⁷.

3. First results of clinical trials with new mTOR pathway inhibitors

The results of several phase 1 clinical trials in advanced cancer (all types) are now available with this new generation of mTOR inhibitors. A phase 1 clinical trial with patients suffering from solid tumors and lymphoma using pan-mTOR inhibitor AZD8055⁸⁸, demonstrated dose limiting toxicities (DLT) of grade 3 with an increase in transaminases and no RECIST objective response⁸⁸. Accordingly, AZD8055 will not be further tested. It is interesting to note that altered liver function was not described with mTOR inhibitors such as temsirolimus or everolimus. It will be important to see if other pan-mTOR inhibitors such as OSI-027, AZD2014, INK128 or CC223 also alter liver function.

Likewise, several dual mTOR-PI3K inhibitors are under evaluation and for 2 of them the phase 1 trial results have already been published. The first, BGT226 induced grade 3 diarrhea in 46% of patients at 125 mg, however, limiting the dose to 100 mg three times weekly resulted in insufficient inhibition of the PI3K pathway⁸⁹. Modeling based on pharmacokinetic data predicted that BGT226 dose of >4000 mg/day would be required to achieve efficacious plasma exposure, exceeding the safety dose range. The second, SF1126 is composed of the pan-PI3K and mTOR inhibitor LY294002 conjugated to an RGDS-targeting peptide to increase binding to integrins expressed on the tumor vasculature. In phase 1, SF1126 did not reach the maximum tolerated dose with a single dose limiting toxicity grade 3 diarrhea, reduced p-AKT and increased apoptosis⁹⁰. Further studies are planned in combination with rituximab in CD20+ B-cell malignancies. Other compounds such as BEZ235, XL765 (also known as SAR245409), GDC-0980, PF-04691502 and PF-05212384 (also known as PKI-587) have completed phase 1 trials but data are not yet published.

Several new compounds have been designed to selectively inhibit PI3K with pan-PI3K targeting all class IA PI3Ks (BKM120, PX-866, XL147, GDC-0941, BAY80-6946, GSK2126458, CH5132799 and ATU027) or PI3K isoform-specific inhibitors (CAL101, BYL719, GSK2636771 and AZD6482). The phase 1 study with BKM120⁹¹ has shown hyperglycemia, skin rash and mood alteration as drug limiting toxicities. Once mood alterations were identified, selective serotonin reuptake inhibitors were prescribed, and no further mood alteration greater than grade 1 was seen. In the expansion study with the MTD (100 mg/d), most frequent grade 3/4 adverse effects (AEs) were transaminase increase (9.1%), asthenia (7.6%) and rash (6.1%). Anti-tumor activity is encouraging with 3 RECIST partial responses. A phase 1 study has also been completed for another compound (PX-866⁹²) showing that dose limiting toxicities consisted of grade 3 diarrhea and elevated AST. No RECIST response was seen in this trial. New trials in association with other drugs are currently ongoing.

Finally, some compounds directly inhibit AKT with different mechanisms of action (plasma membrane disrupting agent, ATP-competitive and allosteric AKT inhibitors). Perifosine, a plasma membrane disrupting agent, has been extensively studied in phase I, II and even III trials^{93, 94}. Despite promising first results with good tolerance and interesting anti-tumor activity, the latest phase III trial in metastatic refractory colorectal cancer was disappointing with no difference of overall survival or progression-free survival for perifosine plus capecitabine compared to capecitabine alone. In a small phase 1 trial with patient selection based on p-AKT positive tumor, triciribine phosphate monohydrate⁹⁵ has shown no dose-limiting toxicities but a modest decrease of p-AKT after treatment. Response rates were not reported. Finally MK-2206, an allosteric AKT inhibitor, recently showed DLTs as mainly grade 3/4 skin rash⁹⁶. In this trial no objective response was seen, however, tumor shrinkage was noted without reaching the level of the RECIST partial response. Of note, 9 patients have had paired tumor biopsy at baseline and at day 15. All of them demonstrated a decrease of p-AKT. Multiple drug combination and weekly schedule are now being tested to increase anti-tumor activity and minimize drug toxicity. Other compounds such as GSK690693, RX0201 (AKT antisense), PBI-05204, GSK2110183, GSK2141795, RG7440, GDC0068 and AZD5363 are under investigation or have completed phase 1 trials but the results are not yet published.

4. Perspective on mTOR inhibitors in HCC therapy

HCC is a very complex and heterogenous disease and progress in treatment of advanced HCC will likely occur slowly. One of the advantages of the first generation inhibitors (rapalogs) is the high specificity, the clinical approval and the comparatively few side-effects. Even if single agent therapies do not show the expected efficacy as has been revealed by the recent results from the EVOLVE study, rapalogs are still very attractive for use with other drugs. Moreover, sirolimus will most likely become first line treatment for HCC patients after liver transplantation.

With regard to current available second generation mTOR inhibitors we can anticipate limitations in their use in HCC patients, although they are more effective than first generation inhibitors in preclinical studies. First, for patients with impaired liver function, the increase in transaminase is troublesome, as observed in several clinical trials with the mTOR inhibitor AZD8055 but also with PI3K/mTOR dual inhibitor BGT226. Second, the performed phase 1 clinical trials demonstrated modest monotherapy efficacy, which was related to adverse events limiting dose escalation, but also related to a cytostatic effect more than a cytolytic effect of these drugs. Therefore, improved efficacy may mainly be reached in combination with other drugs. Third, new generation mTOR inhibitors may stimulate autophagy more than first generation inhibitors, and may consequently cause tumor cell protection against chemotherapy-induced death.

A strong focus should also be directed towards the discovery and validation of biomarkers which predict tumor-response after therapy. Biomarkers for mTOR inhibitor efficacy have been evaluated in preclinical, but also in clinical studies^{97, 98}. Among others, these biomarkers include inactivation of PTEN, activating mutations of PI3KCA, expression levels of pS6, pS6K, S6K, and pAkt. Interestingly, a study demonstrated that human cancer cell lines carrying PI3KCA mutations were responsive to everolimus, except when KRAS mutations occurred concomitantly⁹⁹. Similar results were obtained in a study with colon cancer cell lines¹⁰⁰, and mTOR resistance was also shown in ovarian cancer cell lines, which overexpressed the apoptosis-inhibitory protein Bcl2¹⁰¹. Novel technologies such as next-generation sequencing may further path the way for the identification of new biomarkers, as has been shown in a recent study with bladder cancer patients, in which response to everolimus was clearly more effective in patients with a somatic mutation in the TSC1 complex¹⁰². However, none of these markers have been confirmed in clinical trials. It is thus currently difficult to recommend any of these biomarkers for patient selection in clinical trials. In addition, it has to be emphasized, that substantial progress in the identification of biomarkers for HCC treatment is unlikely to occur, if tumor tissue is not taken before and after chemo-therapy by means of a liver biopsy. Acquiring tumor tissue should therefore be mandatory for any phase II or phase III clinical study and will further help to develop novel therapies.

In spite of these drawbacks, mTOR inhibitors together with AKT and PI3K inhibitors still remain an attractive and promising therapeutic option for the treatment of HCC and ongoing as well as future clinical studies will reveal if they can be used for the therapy of this devastating disease.

Key point box.

• -The mTOR pathway is upregulated in 40-50% of HCC cases, is related to bad prognosis and contributes to sorafenib resistance in HCC. Preclinical in vitro and in vivo studies showed a reduction of HCC tumor growth by mTOR inhibitors.
• -Rapamycin and first generation mTOR inhibitors (rapalogs) are used in clinical trials for advanced HCC but also for adjuvant therapy in HCC patients after transplantation and TACE.
• -Preclinical studies in HCC have shown that second generation mTOR inhibitors are more efficacious than first generation mTOR inhibitors and could synergize with them.
• -Phase 1 clinical trials with second generation mTOR inhibitors in various solid cancers revealed dose limiting toxicities such as diarrhea, skin rash and increase of transaminase
• -Predictive biomarkers for efficacy of mTOR inhibitors have been identified in preclinical studies but not yet confirmed in clinic studies.