Thoughts on investigational hedgehog pathway inhibitors for the treatment of cancer

KEYWORDS:

• Hedgehog pathway
• inhibitors
• drug development

1. Introduction – oncogenic hedgehog signaling

Intensive investigations of embryonic development and differentiation pathways in different species established several ‘classical’ pathways such as WNT, NOTCH, and hedgehog (Hh) signaling [1]. The increasing understanding of molecular carcinogenesis revealed that tumor stem cell signaling and reactivation of embryonic differentiation pathways like Hh can be oncogenic drivers for solid and hematologic tumors. From development, differentiation, tissue homeostasis, and maintenance as well as carcinogenesis, the next essential step is to develop highly specific drugs to reverse such uncontrolled pathway activation and therefore to prevent cancer initiation and, especially, progression and metastasis. Numerous small-molecule inhibitors targeting Hh signaling have been successfully developed pre-clinically, and two are currently approved by the US Food and Drug Administration – sonidegib (LDE225) and vismodegib (GDC-0449) [2]. Targeted inhibitors of embryonic signaling pathways like Hh could significantly contribute to improved cancer therapies as single agents or especially in combination with standard-of-care chemotherapy or other targeted therapies. Success of early clinical studies was so far limited by inadequate patient selection, lack of predictive biomarkers, and development of resistance. Increasing efforts need to be put into identifying such biomarkers using novel technologies like next-generation sequencing, RNA-based patient selection, and comprehensive use of liquid biopsies.

Here, we will discuss the review ‘Emerging from Their Burrow: Hedgehog Pathway Inhibitors for Cancer’ by Gan and Jimeno in this issue of Expert Opinion On Investigational Drugs [3] and develop thoughts on the appropriate next steps in clinical establishment of Hh inhibitors. The Hh pathway has been established as an oncogenic driver in human carcinogenesis by mediating epithelial-to-mesenchymal transition via interaction of the tumor cell with its microenvironment and gaining of a stem-cell-like phenotype in addition to promoting other hallmarks of cancer. Increasing efforts have recently been started to inhibit Hh-associated molecules at different stages of its signaling cascades [1,4].

As summarized in Figure 1(a), Hh signaling involves binding of the ligands Sonic (Shh), Indian (Ihh), or Desert (Dhh) Hh to the cell surface receptor Patched (Ptc). This relieves the inhibitory effect on the downstream signal transducer smoothened (SMO) and activates nuclear transcription factors of the Gli family, thus inducing cognate Hh target genes linked to proliferation, survival, and angiogenesis [2]. Typical mechanisms of dys- and hyper-regulated activated Hh pathways are mutations of the PTCH1 (loss-of-function) and SMO (gain-of-function) genes which were observed in basal cell carcinoma of skin, medulloblastoma, and, to a lesser extent, rhabdomyosarcoma. Importantly, the Hh pathway has a canonical and a different non-canonical activation mode: the latter includes stress and inflammatory response mechanisms as well as activation of other pathways such as TGF-β, KRAS-MAPK/ERK, PI3K-AKT, IGF, TNF-α-induced mTOR/S6K1, or inactivation of the hSNF5 PI3K/AKT/mTOR/S6K1 axis. Of note, such cross-pathway activation could bypass inhibitory mechanisms of other anticancer drugs [4]. The cross talk between the canonical versus the non-canonical Hh signaling likely suggests alternative activation (or escape) mechanisms which (cancer) cells may exploit when faced with cytotoxic or inflammatory injury.

Figure 1. Hedgehog pathway as potential therapeutic target.

Inhibitors of Smoothened (Smo) and arsenic trioxide (ATO) are currently being evaluated in clinical trials; see text for details. a) Overview of the Hedgehog pathway. * indicate drugs which were shown to inhibit the vismodegib-resistant Smo mutant (D473H). b) Aspects for optimized clinical trial design. Abbreviations: Bx, biopsy; DHh, Desert hedgehog; Histo, histology; IHC, immunohistochemistry; IHh, Indian hedgehog; ISH, in situ hybridization; NGS, next generation sequencing; Ptc, Patched; SHh, Sonic hedgehog; SKN, Skinny hedgehog; SUFU, suppressor of fused; Tx, therapy. Based on [1–3].

2. Targeting Hh – current approaches

Over the last decades, cancer treatment strategies with several specific targets have been developed to achieve a breakthrough in cancer therapy. One milestone was the development of the targeted tyrosine kinase inhibitor imatinib for the treatment of chronic myelogenous leukemia in the 1990s. However, the initial great expectations were dampened by the appearance of drug resistance with relapse of the disease after initial therapy success. To overcome this problem, the combination of multiple drugs has been suggested similar to the use of a ‘drug cocktail’ in the successful treatment of AIDS. Current approaches for cancer therapy are to administer different agents consecutively, beginning with ‘first-line’ therapy and changing the drug to a ‘second-line’ therapy in the case of tumor recurrence.

As presented in detail by Gan et al. in the review ‘Emerging from Their Burrow: Hedgehog Pathway Inhibitors for Cancer,’ several Hh inhibitors targeting Shh, SMO, or Gli are currently being evaluated (see also Figure 1(a) and Table 1). The first pre-clinical in vitro and in vivo analyses of these drugs based on investigations of unbiasedly selected tumor cell lines and xenografts seem to be very convincing – as also shown by own experiments [5].

Table 1. Overview of current anti-hedgehog-based clinical trials.

Smoothened inhibitors	Number of clinical trials
	All	I	II	III	IV	Recr.	Comp.	Term.
Vismodegib	71	20	47		2	37	27	7
Glasdegib	11	7	4			6	5	0
Saridegib	7	4	3			1	6	0
Sonidegib	37	22	14			19	12	3
Taladegib	7	7	2			4	2	1
TAK-441	1	1				0	1	0

Source: clinicaltrials.gov (by November 2016).

Comp.: completed; recr.: recruiting; term.: terminated.

Based on this rationale, translational clinical trials using Hh inhibitors alone or in combination with other clinically established drugs were initiated. From a larger number of drugs developed pre-clinically, seven inhibitors specifically targeting Hh signaling were recently/currently used in clinical development studies: vismodegib, sonidegib, glasdegib, saridegib, taladegib, TAK-441, and arsenic trioxide (ATO; Table 1). It is noteworthy that besides ATO, specific inhibitors of Gli transcription factors were not investigated clinically until now, and additional agents such as itraconazole were used in the clinical setting as Hh-antagonizing agents [2,4]. Potential indications for Hh inhibitors in early clinical studies included (i) solid tumors such as classical Hh-driven basal cell carcinoma of the skin and medulloblastoma as well as prostate, pancreatic, gastroesophageal, hepatocellular, and breast cancers or keratocystic odontogenic tumors; (ii) mesenchymal tumors (chondrosarcoma and meningiomas); and (iii) different hematopoietic malignancies including acute myeloid leukemia, multiple myeloma, myelofibrosis, myelodysplastic syndrome, and chronic myelomonocytic leukemia [1,3]. Results of these studies were heterogeneous due to numerous genetic and epigenetic alterations which are commonly found in advanced solid malignancies – the common patient group within oncology phase 1 studies.

3. Current challenges in antitumor Hh targeting

As discussed in detail in the review ‘Emerging from Their Burrow: Hedgehog Pathway Inhibitors for Cancer,’ current experience with Hh-targeting drugs is not fully convincing. While pre-clinical models suggest efficient antitumor effects against various tumor types (e.g. for Gli-targeting GANT drugs [5]) and promising initial data exist for e.g. BCC and medulloblastoma [6,7], especially with vismodegib, several studies reported acquired resistance towards this Smo inhibitor [8,9]. Additionally, evidence for Hh drugs (saridegib) combined with chemotherapy showed worse clinical data compared to conventional therapies [10]. Results of the studies combining vismodegib with standard chemotherapy in unselected patient populations have been disappointing. This trial failed to show that inhibition of Hh-signaling networks between tumor cells and stromal cells might have superior clinical antitumor activity in combination with standard-of-care chemotherapy compared to chemotherapy plus placebo. Furthermore, mRNA expression analysis of Hh ligands, SMO, or PTCH1 mRNA in tumor had no predictive value [11].

Single-drug therapy may provide a competitive advantage to drug-resistant cancer cell mutants, while the switch to a second-line therapy might in turn implicate the threat of acquiring resistance against the second drug (double resistant mutant). Combination of therapy agents may therefore increase the chance of therapy success by eliminating single resistance and by lowering the emerging of a double mutated cell. However, recent results from combination regimes for locally advanced rectal cancer (e.g. STAR-01, ACCORD12/0405, NSABPR-04, and PETACC 6 trials) do not support the suggested beneficial effect of such chemotherapy combinations [12]. Concomitant local ablative therapies (e.g. hypofractionated or stereotactic body radiotherapy) may improve and/or prolong local/systemic tumor control – nevertheless, combination of chemotherapies with specific Hh inhibitors is worth further studies to elucidate their potential.

Taken together (reviewed in [1,2]), current challenges include (i) improving of the pharmacokinetic profile (to increase intratumoral drug concentration), (ii) prevention of non-canonical Hh activation (at Gli transcript factor level), and (iii) detection and managing of possible drug-induced mutations.

3.1. How to improve the Hh drug efficiency?

In light of the current experience with small-molecule Hh-targeting drugs (as reviewed in [1,3]), the following aspects should be considered in future trial design and implementation for successful translation into clinics (see also Figure 1(b)):

• Patient stratification: Selection of potentially responsive patients eligible for Hh-based therapies by identification and validation of predictive biomarkers with a stronger focus on liquid biopsies and the tumor environment.

• Therapy monitoring: Surveillance of pharmacodynamic biomarkers and key mutations indicating development of resistance as well as the therapeutic effect – preferably via liquid biopsies or non-invasive imaging technologies.

• Rational combinations: Combining Hh-targeting drugs with conventional chemotherapy as well as other targeted drugs to simultaneously target con-canonical compensatory Hh signal transduction.

• Development of non-redundant drugs targeting Hh components: Drugs with alternative molecular targets (other pathway components or molecular docking points) may help overcoming development of drug resistance. For example, SMO inhibitors such as TAK-441, taladegib, and MRT-92 have shown activity also against SMO mutants conferring resistance towards vismodegib.

4. Expert opinion

Numerous small-molecule inhibitors of Hh signaling have been developed within the last 10–15 years. While some of these compounds show sufficient efficacy and safety to warrant early clinical development in humans, the final proof of concept that interfering with embryonic differentiation and tumor stroma cross talk leads to improved overall response rates and longer progression-free or overall survival compared to standard of care is still missing. While this concept is well accepted for hematologic tumors, it is still not completely clear which implications the inhibition of stem cell-like pathways will have in solid tumor therapy. Several compounds show only weak monotherapy activity especially in in vitro systems. The activity in murine models also needs to be carefully evaluated as, e.g. cancer cell lines being xenografted to nude mice may have a different proportion of cancer stem-like cells than primary human tumors. Although patient-derived xenografts could improve the quality of these experiments, these models still lack the complex interplay between tumor cells, immune system, and (human) stroma and matrix components in mice.

Many of such innovative and novel potential therapeutic pathways fail to translate successfully into the human setting. Despite lacking (monotherapy) efficacy, significant toxicity can be induced by inhibition of these non-redundant signaling pathways, which is in contrast to tyrosine kinase inhibitor signaling. Thus, inhibitors of, e.g. tankyrase signaling were terminated due to severe gastrointestinal toxicity [13]. It is currently also unclear to what extent Hh signaling is an oncogenic driver in advanced solid tumors. As these drugs aim to target slow growing stem cell-like populations, resistance development might occur fast and thus counteract the effect of such targeted agents.

This provides the option for rational drug combinations that could overcome resistance against Hh inhibitors, e.g. by using already approved PI3K or AKT inhibitors. Interesting new combination options might come up with the broader use of immune checkpoint inhibitors (e.g. programmed cell death protein 1 (PD-1, CD279), its ligand (PD-L1, CD274), and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4)). Clinical results of these antibodies are encouraging although it is still not completely understood what biomarker predicts the long-term response these drugs could induce in various indications. Preliminary preclinical data indicate that stem cell targeting, e.g. with WNT inhibitors, could turn tumors from ‘cold’ to ‘hot’ or activate T-cell maturation and migration into the tumor spot and could also provide an opportunity for otherwise only fairly active Hh inhibitors in humans.

A key task for the successful development of such drugs is the identification of the appropriate patient group to obtain optimal responses. While classical histology-based patient stratification might not be sufficient, more efforts need to be put into identification and validation of biomarkers predictive of therapeutic response towards Hh-based drugs. The existence of such a predictive biomarker significantly increases the overall probability of technical success from bench to launch [14] (see also https://www.bio.org/bio-industry-analysis-published-reports) but is currently considered low for Hh targeting agents as even in classically Hh-driven malignancies like medulloblastoma only 15–30% actionable mutations were found [15]. Additionally, novel methods in addition to conventional immunohistochemistry need to be applied to clinical trials. Here, especially next-generation sequencing approaches hold great promises to obtain multi-layer genetic and functional data from tissue samples. A further step would be to implement this technology also to circulating tumor (stem) cells. Other RNA based-technologies such as nanostring quantification or RNA in situ hybridization offer different advantages like the spatial resolution of expression are now already used in clinical trials to improve the selection of eligible patients for novel (Hh) targeted drugs.

Future challenges in this area will also include the changing perspectives of regulatory authorities, health technology agencies, payers, and patient advocacy groups. Repeated retesting of patients will also pose additional challenges on the quality and procedures of taking solid tumor biopsies, and thus, the use of liquid biopsies will become more important. In summary, additional careful and technically sophisticated efforts are required to allow for successful translation of the promising pre-clinical approaches to clinically applicable Hh-based anticancer therapies.

Declaration of interest

M Ocker is an employee and shareholder of Bayer Pharma AG, Berlin, Germany. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Additional information

Funding

T Kiesslich was supported by research grants of the Anniversary Fund of the Oesterreichische Nationalbank (OeNB, grant No. 12677, 14842) and the research fund of the Paracelsus Medical University Salzburg (grant No. 08/07/037, A- 1 21 021 006-KlE).