The use of hedgehog antagonists in cancer therapy: a comparison of clinical outcomes and gene expression analyses

ABSTRACT

Hedgehog (HH) signaling, a critical developmental pathway, has been implicated in cancer initiation and progression. With vismodegib and sonidegib having been approved for clinical use, increasing numbers of HH inhibitors alone and in combination with chemotherapies are in clinical trials. Here we highlight the clinical research on HH antagonists and the genetics of response to these compounds in human cancers. Selectivity of HH inhibitors, determined by decreased pathway transcriptional activity, has been demonstrated in many clinical trials. Patients with advanced/metastatic basal cell carcinoma have benefited the most, whereas HH antagonists did little to improve survival rates in other cancers. Correlation between clinical response and HH gene expression vary among different cancer types. Predicting response and resistance to HH inhibitors presents a challenge and continues to remain an important area of research. New approaches combine standard of care chemotherapies and molecularly targeted therapies to increase the clinical utility of HH inhibitors.

Introduction

The Hedgehog (HH) signaling pathway is vital to the process of embryogenesis, including cell growth, differentiation and overall embryonic development. While essential for regulating development, abnormal activation of HH signaling and mutations in HH signaling genes have been linked to the development and progression of a number of human cancers, including basal cell carcinoma,^1,2 medulloblastoma,³ breast cancer,⁴ pancreatic cancer,^5,6 and many more.^7–9 Based on the relationship between initiation and progression of cancer and Hedgehog signaling, targeted therapeutics have been developed to inhibit the pathway and its downstream effects. First-in-class HH inhibitors currently on the market include vismodegib (Genentech) and sonidegib (Sun Pharma). In this review, we will discuss the clinical research of these, as well as other in-development Hedgehog antagonists being used in a variety of human cancers in relation to their distinct genetic effects.

The hedgehog pathway

The HH pathway is comprised of a family of secreted morphogens that were first described in Drosophila in the early 1980s, and in mammals in the 1990s, as playing a crucial role in embryonic development, particularly organ and limb patterning.^10–13 Nusslein-Volhard and Wieschaus¹⁰ were the first to name the hedgehog gene after noticing that when this gene is mutated, denticles (spiked cuticle that normally decorates only the anterior portion of fly body segments) were prevalent throughout the entire body (both anterior and posterior portions) of newly hatched larva. This continuous lawn of denticles suggested the spines of a hedgehog.

Vertebrate HH genes were first reported in 1993 when a collaborative effort involving three groups^14–16 discovered that, unlike the fruit fly, which has a single hh gene, there are three related mammalian HH genes: Desert HH, Indian HH, and Sonic HH. Sonic HH, named after the title character in the popular video game series, is the best characterized of the three HH ligands as it appears to play a substantial role in the formation of several tissues and organs.¹⁷ In canonical HH signaling, these HH ligands are secreted by cells and interact with the 12-span transmembrane receptor Patched (PTCH), either in an autocrine or paracrine manner, thereby relieving PTCH-mediated inhibition of Smoothened (SMO), a 7-span transmembrane protein.^18–20

In the absence of HH ligand, PTCH inhibits signal transduction through SMO, resulting in the cytoplasmic sequestration of the Glioma-associated oncogene (Gli) family of transcription factors (Gli-1, 2, 3).^21,22 While sequestered, Gli is phosphorylated and ultimately marked for proteasomal degradation.^23–25 At this point, the ultimate fates of the three Gli transcription factors diverge somewhat. Proteolysis of Gli-1 by the proteasome is a complete process whereas degradation of Gli-2 and Gli-3 is only partial. Gli-2 and Gli-3, unlike Gli-1, have activator and repressor domains. Upon proteolysis, Gli-2 and Gli-3 are cleaved into smaller transcriptional repressor fragments (Gli-2 R and Gli-3 R) with Gli-2 R being a much weaker repressor than Gli-3 R, which localize to the cell nucleus where it binds Gli-response elements to prevent HH signal transduction.^25,26 The degradation of Gli is inhibited in the presence of HH ligand (Figure 1). When HH binds to PTCH, SMO becomes activated resulting in a signaling cascade that ultimately leads to the translocation of the full-length, activator forms of Gli into the nucleus where they transcribe HH target genes including PTCH1 and GLI1, thereby providing a mechanism whereby HH signaling is tightly regulated.²⁷

Figure 1.

The hedgehog signaling pathway. In the presence of hedgehog ligand, repression of the SMO receptor by the PTCH receptor is relieved, allowing Gli transcription factors to translocate to the nucleus and transcribe hedgehog target genes. Pharmacologic inhibitors of SMO and Gli (shown in boxes) have been developed for cancer therapy

In contrast to the classic, canonical ligand-Patched-Smo regulated Hedgehog signaling; alternative, non-canonical activation of HH signaling has been shown in both the physiological and pathophysiological effects attributed to the pathway. Models of non-canonical activation of HH signaling include PTCH-dependent, Smo-independent (Type I) and Smo-dependent, PTCH-independent (Type II) signaling, as well as GLI activation driven by other pathways.^28–30 Of particular interest related to the hedgehog inhibitors discussed here is PTCH-dependent, Smo-independent signaling and Smo-independent GLI activa-tion, as both would be refractory to Smo inhibition and may be difficult to distinguish from canonical HH signaling using traditional HH pathway assessments.

Hedgehog pathway inhibition and cancer therapy

In the last decade, several in vitro and in vivo studies have implicated aberrant HH signaling in a variety of cancers, either due to loss-of-function mutations in PTCH1 or gain-of-function mutations in SMO. These discoveries have led to ongoing research devoted to identifying and testing inhibitors of the HH pathway. Cyclopamine, a naturally occurring steroidal alkaloid, represents the first member of a class of small molecule compounds that selectively target SMO.^31,32 This compound inactivates HH transcriptional activity by directly binding to SMO’s heptahelical bundle and inducing a conformational change similar to that induced by PTCH.^33,34 Cyclopamine is so named due to its ability to induce holoprosencephaly (i.e., cyclopia) in the developing fetuses of pregnant sheep who had the misfortune of ingesting Veratrum californicum, the plant from which this compound is derived.³⁵ Cyclopamine was first used as an anti-cancer agent to inhibit the proliferation of brain cancer cells in vitro.³⁶ However, the clinical use of cyclopamine is hindered due to sub-optimal chemical stability, poor solubility in aqueous solutions, and a lack of HH specificity.^37,38

Using cyclopamine as a “proof of concept,” SMO antagonists that are structurally distinct from cyclopamine and more selective for SMO have been developed by Genentech (GDC-0449/vismodegib),³⁹ Novartis (LDE-225/sonidegib and LEQ-506), ^40,41 Pfizer (PF-04449913/glasdegib),⁴² Millennium (TAK-441),⁴³ Bristol-Myers Squibb (BMS-833923)⁴⁴ and Eli Lilly (LY2940680/taladegib)^45,46 pharmaceutical companies. In addition, Infinity Pharmaceuticals has taken cyclopamine and chemically modified it to make patidegib (IPI-926/saridegib), which is more selective for SMO and improves bioavailability.^47–49 All of these compounds have been or are currently in clinical trials, either as single agents or in combination with other chemotherapy agents, for the treatment of a variety of cancers. Vismodegib and sonidegib were FDA-approved for the treatment of basal cell carcinoma in 2012 and 2015, respectively (see Table 1 for a list of hedgehog antagonists currently in clinical development). Response to SMO antagonists varies among different cancer types and the molecular basis for this differential response remains an active area of research. By elucidating this mechanism, it may be possible to identify cancer patients most likely to respond to pharmacological HH inhibition based on their tumor’s genetic or expression profile.

Table 1.

Smoothened antagonists currently under clinical investigation

					Route of
Compound	Chemical	Generic Name	Brand Name	Developer	Administration	Clinical Trials	FDA-approved
Cyclopamine	Natural	n/a	n/a	n/a	n/a	None	No
GDC-0449	Synthetic	vismodegib	Erivedge	Genentech	Oral	Phase I,II	Yes-2012
LDE-225	Synthetic	sonidegib	Odomzo	Novartis	Oral	Phase I,II	Yes-2015
IPI-9261	Semi-Synthetic	saridegib	n/a	Infinity	Oral	Phase I,II	No
PF-04449913	Synthetic	glasdegib	n/a	Pfizer	Oral	Phase I,II	No
TAK-441	Synthetic	n/a	n/a	Millennium	Oral	Phase I	No
BMS-833923	Synthetic	n/a	n/a	Bristol-Myers Squibb	Oral	Phase I,II	No
LY2940680	Synthetic	taladegib	n/a	Eli Lilly	Oral	Phase I,II	No
LEQ-5062	Synthetic	n/a	n/a	Novartis	Oral	Phase I	No

Clinical trials and genetic correlates

Primary literature and clinical trial registration data were reviewed to provide information on human clinical trials that examined cancer response to HH antagonism. All cancer types (both hematologic and non-hematologic/solid) and all clinical trial phases (I, II, III) were examined. Compounds designed to selectively target SMO (vismodegib, sonidegib, saridegib, etc.) were included, whereas compounds that inhibit HH signaling as part of an off-target effect were excluded from the selection (e.g., itraconazole). Studies must have included a genetic analysis component in which gene expression from available biopsies was compared to SMO inhibitor response (either pre- or post-treatment). Studies examining only gene expression or only drug/clinical response were excluded. Clinicaltrials.gov was used to identify SMO inhibitor compounds that are currently in clinical trials, as well as ongoing, and recently completed clinical trials; but the results of which have not yet been reported or published.

Fifteen clinical trials were located in PubMed using the inclusion/exclusion criteria described above (Table 2). Eight of the trials focused on vismodegib,^50,51,57-62 two on sonidegib,^52,63 two on patidegib (IPI-926), ^53,54 one on PF-04449913 (glasdegib),⁵⁵ one on TAK-441,⁵⁶ and one on LY2940680 (taladegib).⁴⁶ In addition, clinicaltrials.gov showed that two other SMO inhibitors, BMS-833923 and LEQ-506, are currently in clinical trials. Many published Phase I and II clinical trials demonstrated that SMO antagonists (vismodegib, sonidegib, patidegib, glasdegib, TAK-441) are selective for their target, as demonstrated by a decrease in GLI1 expression in patient tumor samples using mRNA or protein levels.^46,52-59 These trials also found that SMO antagonists are well tolerated, with the side effects of dysgeusia, fatigue, muscle spasms and alopecia belonging to a drug class effect. Dose finding studies in solid tumors identified 150 mg/day vismodegib, 800 mg/day sonidegib, 160 mg/day patidegib, 100 mg/day glasdegib and 1600 mg/day TAK-441 as safe and effective doses.

Table 2.

Summary of smoothened antagonist clinical trials with genetic analysis

Study by Author	Year	Cancer Type	Phase	Design	Treatment	Primary Endpoint	Target Inhibition	Ref #
Von Hoff, DDa	2009	Solid tumors	I	Open-label	Vismodegib	Safety	↓ GLI1 pt, correlates w/co	50
LoRusso, PMa	2011	Solid tumors	I	Open-label	Vismodegib	Safety	↓ GLI1 pt, correlates w/co	51
Rodon, J	2014	Solid tumors	I	Open-label	Sonidegib	Safety	↓ GLI1 pt, no correlation w/co	52
Jimeno, A	2013	Solid tumors	I	Open-label	Saridegib	Safety	↓ GLI1 pt, no correlation w/co	53
Bowles, DW	2016	SCC	I	Open-label	Saridegib + cetuximab	Safety	↓ HH pathway pt, correlates w/co	54
Wagner, AJ	2015	Solid tumors	I	Open-label	Glasdegib	Safety	↓ GLI1 pt, no correlation w/co	55
Goldman, J	2015	Solid tumors	I	Open-label	TAK-441	Safety	↓ GLI1 pt, no correlation w/co	56
Bendell, J.	2018	BCC	I	Open-label	Taladegib	Recommended Phase II	↓ GLI1 pt, no correlation w/co	46
Tang, JYb	2012	BCC	II	R, DB	Vismodegib	Reduced incidence of new BCC	↓ GLI1 pt, correlates w/co	57
Kaye, SB	2012	OC	II	Randomized	Vismodegib	PFS	No correlation	58
Berlin, J	2012	CRC	II	R, DB	Vismodegib + SOC	PFS	No correlation	59
Italiano, A	2013	CHS	II	Open-label	Vismodegib	Patients with non-PD at 6 months	No correlation	60
Kim, EJ	2014	PANC	II	Open-label	Vismodegib + SOC	PFS	↓ GLI1 pt, no correlation w/co	61
Robinson, GW	2015	MB	II	R, DB	Vismodegib	PFS	Correlation between baseline gene expression and co	62
Migden, MRb	2015	BCC	II	Open-label	Sonidegib	Objective response rate	↓ GLI1 pt, correlates w/co	63
Maughan	2016	PC	II	Open-label	Vismodegib	Pharmacodynamics	↓ GLI1 pt, no correlation w/co	64

^aBoth studies are based on the same clinical trial

^bServed as the basis for FDA approval

Abbreviations: BCC, basal cell carcinoma; CRC, colorectal cancer; CHS, chondrosacroma; OC, ovarian cancer; PANC, pancreatic cancer; PC, prostate cancer; SOC, standard of care; MB, medulloblastoma; SCC, squamous cell carcinoma; PD, progressive disease; R, DB, random, double blind; PFS, progression-free survival; pt, post-treatment; co, clinical outcome

Among the different cancer types examined, patients with locally advanced or metastatic BCC benefited the most from Smo antagonist therapy and, to a lesser extent, those with medulloblastoma. In Phase I clinical trials including patients with advanced tumors, vismodegib, sonidegib, and patidegib demonstrated objective or partial response in BCC patients compared to other advanced tumors (e.g. pancreatic, colorectal, lung, breast, and ovarian) .^51–53 Later clinical trials for metastatic colorectal, ovarian, pancreatic, prostate, and non-small cell lung carcinoma showed vismodegib combined with chemotherapies did not provide clinical benefit compared to chemotherapy alone.^{58,59,61,64-67} In BCC patients, both vismodegib and sonidegib demonstrated objective disease control (with stable disease or tumor regression) in a majority (>90%) of patients.^{51–53,57,63} This clinical response was accompanied by a substantial decrease in GLI1 expression (>90%) in available biopsies post-treatment compared to pre-treatment. Progression-free survival was found to be significantly longer in medulloblastoma patients treated with vismodegib whose tumors expressed activating mutations in the HH pathway compared to those whose tumors did not have HH activation.⁶² Similarly, longer progression-free survival (>150 days) was associated with HH pathway downregulation in patients with squamous cell carcinoma after treatment with patidegib, whereas shorter progression-free survival was seen in patients with little or no change in HH pathway gene expression.⁵⁴ Taken together, these studies suggest that gene expression (either before or after treatment) can correlate with HH antagonist clinical outcomes.

Conversely, other solid tumors (e.g. pancreatic, colorectal, lung) treated with SMO antagonists did not demonstrate a correlation between clinical response and genetic changes. Several of the clinical trials found that GLI1 mRNA expression was significantly decreased in available post-treatment biopsies (from 57–100% of patients), but there was no meaningful clinical response (e.g. progression-free survival) in these patients.^{52,53,55,56,61,64} For example, 57% of patients with metastatic, castration-resistant prostate cancer showed a reduction in expression of GLI1, GLI2 and PTCH1 with vismodegib treatment, but no clinical response.⁶⁴ Of note, combination of vismodegib with standard of care (SOC) chemotherapy agents did not improve survival in patients with colorectal or pancreatic cancer compared to SOC chemotherapy alone.^59,61 In addition, other clinical trials demonstrated that HH gene expression in biopsies taken from cancer patients before treatment (i.e., baseline expression) did not correlate with clinical response to SMO antagonists and, therefore, could not predict which patients might respond to this therapy.^58–60 Glasdegib (PF-04449913) combined with chemotherapy in patients with acute myeloid leukemia and myelodysplastic syndrome did achieve a more than 2-fold change in expression in SMO and PTCH2,^67–69 but failed to demonstrate significance between the molecular expression and clinical responsiveness. Recently, taladegib (LY2940680) began a Phase I clinical trials for advanced basal cell carcinoma (BCC), colon adenocarcinoma, and other advanced solid tumors.⁴⁶ This study included both patients that had received a previous hedgehog inhibitor and treatment-naïve patients to determine if patients with prior hedgehog inhibitor exposure could be differentiated from treatment-naïve patients based on clinical response. Clinical response was observed for both treatment-naïve patients (68.8%) and those that had received a previous hedgehog inhibitor (35.5%); however, no statistically significant differences were observed amongst groups. Overall, GLI1 expression was inhibited by 92.3% compared to normal skin; however, % inhibition did not correlate with clinical response. Baseline Indian HH, Sonic HH, SMO, and GLI1 expression was evaluated in ovarian cancer patients,⁵⁸ Indian HH, Sonic HH, SMO, PTCH1 and GLI1 in colorectal cancer patients⁶¹ and Sonic HH, SMO, GLI2, and GLI3 in chondrosarcoma patients.⁶⁰ These molecular markers failed to correlate with clinical response.

Many of these solid tumors lack mutations in PTCH1 or SMO. These mutations activate HH signaling and are prevalent in tumors that respond to HH inhibitor treatments with corresponding decreases in GLI1.⁵¹ However, solid tumors lacking PTCH or SMO mutations may have increased expression of HH ligands or other components of the pathway (e.g., GLI1), which could be indicative of canonical or non-canonical signaling. So, while Smo antagonists failed to provide a correlation between clinical response and genetic changes in these solid tumors, non-canonical HH signaling may be important when interpreting these findings. Combinatorial or alternative therapies may target both canonical and non-canonical mechanisms of HH signaling and be a viable option for future studies

New approaches and future directions

Taken together, these clinical trials demonstrate that SMO antagonists are selective for their target in human cancers, can effectively reduce GLI1 expression, are well tolerated, but still have variable clinical efficacy. Clinical response to these compounds varied among different cancer types, with some cancers demonstrating a correlation between clinical outcomes and baseline HH gene expression or gene expression changes after treatment, whereas other cancers did not. BCCs, the most common type of skin cancer, appear to respond best to SMO antagonism compared to other cancer types, both in terms of HH pathway inhibition and clinical benefit. A majority of basal cell carcinomas are known to have loss of function mutations in PTCH1 or activating mutations in SMO, making HH signaling in these tumors ligand-independent.² Tumors dependent upon HH ligand signaling (either in an autocrine or paracrine manner) appear to be less sensitive to SMO antagonists as single agents.^58–60 In addition, there have been documented cases of resistance to SMO antagonists due to acquired mutations in the SMO protein that can change drug-binding affinity, diminish, or prevent the compounds from binding to the heptahelical bundle.^70,71 Vismodegib and sonidegib bind to the same drug-binding pocket on the SMO protein, and new therapies are under development to target different domains of SMO.^37,50,71,72 In vitro research has shown that the SMO mutation E518A (a glutamic acid to alanine change at amino acid 518 of SMO) increases the binding affinity for sonidegib, but reduces the affinity for vismogedib.^73,74 SMO mutant D427H results in a disruption of sonidegib binding to SMO as a result of a conformation change in the transmembrane domain of SMO.⁷⁵ This phenomenon is reported in patients as well, new mutations in SMO, not present in the pretreatment tumor, were observed in BCC patients with acquired resistance to vismodegib.⁷⁶ SMO D473H prevents the effective binding of vismodegib and sonidegib, thus failing to inhibit SMO activity with either drug.^49,75 In clinical specimens from BCC patients with resistance to Hedgehog antagonists, 50–69% had identified SMO mutations, which possibly confer resistance.^77–79 Pietrobono and Stecca⁷⁷ report numerous SMO mutations identified in human cancers and their role in drug resistance.⁸⁰ Second generation SMO antagonists are currently under development with the goal of overcoming this resistance.⁴¹ Novartis’ LEQ-506 was developed to overcome the conformation change induced by the D473H mutation in SMO. It has not been fully tested for efficacy, but a Phase I clinical trial has been completed with no results yet available.^75,81 The molecular basis for these resistance mechanisms continues to be studied, and may offer some explanation as to why certain cancers respond to Smo antagonists in both a targeted and clinically meaningful way, while others do not.

The use of SMO antagonists in HH ligand-dependent cancers may have some merit, as evidenced by the study from Kaye et al.⁵⁸ which demonstrated a trend toward increased progression-free survival among ovarian cancer patients treated with vismodegib alone compared to placebo (7.5 months versus 5.8 months, respectively). Perhaps there exists a sub-population of ovarian cancer patients responsive to SMO antagonists that could be identified for future clinical trials, a concept that is currently being applied to other cancer types. A developing trend observed amongst clinical trials is the selection of drug regimen across a multiplicity of cancers based on the cancer or tumor’s molecular signature (basket or umbrella trials).^82–87 Many of these trials include hedgehog inhibitors as a treatment option. Patients with recurrent brain tumors (young adults),⁸² medulloblastomas,⁸⁴ refractory solid tumors, lymphomas, and multiple myeloma,⁸³ and other advanced cancers^85,86 are initially being analyzed for the molecular signature of their diseases, in order to give targeted single or combination therapies, which may include Hedgehog inhibitors. This sets the foundation for a broader examination of patient pharmacogenomics, possibly using cDNA microarrays or RNA sequencing,^54,62 rather than quantifying the expression of a limited number of genes. In addition, changes at the epigenetic or protein level could be evaluated to better understand individual differences in response to SMO antagonists. If a molecular signature of response is identified, it may be used to screen for candidates who would gain the most benefit from HH inhibitor therapy, much the same way pharmacogenetic testing is now done for medications such as clopidogrel (CYP2C19), warfarin (CYP2C9), trastuzumab (HER2/neu) and imatinib (Bcr-abl).

Due to the limited therapeutic utility of SMO antagonists as single agents in cancer types outside of BCC and medulloblastoma, there is now movement toward combining HH inhibition with standard chemotherapy agents (Table 3). While initial published clinical trials examining this concept have yielded underwhelming results,^59,61,88 there are numerous pre-clinical studies that suggest combination therapy could still be a useful strategy, possibly due to crosstalk between ligand-dependent HH signaling and other survival pathways.⁸⁹ According to clinicaltrials.gov, vismodegib, sonidegib, patidegib/IPI-926, glasdegib/PF-04449913, BMS-833923 and taladegib/LY2940680 combined with other cancer therapeutic agents are currently under clinical investigation (Tables 3 and 4). These combination therapies are largely being conducted in patients with non-hematologic/solid tumors (pancreatic, lung and breast) in comparison to hematologic malignancies (acute myeloid leukemia, chronic myeloid leukemia and multiple myeloma). A pilot Phase I study combined cetuximab (an epidermal growth factor receptor antibody) and saridegib in patients with recurrent or metastatic head and neck squamous cell carcinoma (HNSCC).⁵⁴ Patient response correlated with expected changes in EGFR and HH-related genes as measured via RNA sequencing. Favorable patient outcomes correlated with a decrease in HH pathway genes, showing effective targeting of the HH pathway.

Table 3.

Clinical trials examining smoothened antagonists combined FDA approved drugs

Smo Antagonist	Therapy (Target)	Cancer	Identifier
Vismodegib	Ribavirin±decitabine	AML	NCT02073838
Vismodegib	Erlotinib (EGFR)±gemcitabine	Pancreatic	NCT00878163
Vismodegib	Cisplatin+etoposide	SCLC	NCT00887159
Vismodegib	Sirolimus (mTOR)	Pancreatic	NCT01537107
Vismodegib	Pembrolizumab (PD-1)	BCC	NCT02690948
Vismodegib	Temozolomide	Medulloblastoma	NCT01601184
Vismodegib	Nivolumab (PD-1)	BCNS	NCT03767439
Sonidegib	Ribociclib	Brain	NCT03434262
Sonidegib	Azacitidine+decitabine	Leukemia	NCT02129101
Sonidegib	Fluorouracil+leucovorin+oxaliplatin+irinotecan	Pancreatic	NCT01485744
Sonidegib	Gemcitabine+Nab-paclitaxel	Pancreatic	NCT02358161
Sonidegib	Docetaxel+prednisone	Prostate	NCT02182622
Sonidegib	Lenalidomide	MM	NCT02086552
Sonidegib	Nilotinib (tyrosine kinase)	CML	NCT01456676
Sonidegib	Etoposide+cisplatin	SCLC	NCT01579929
Sonidegib	Everolimus (mTOR kinase)	Esophageal	NCT02138929
Sonidegib	Gemcitabine	Pancreatic	NCT01487785
Sonidegib	Docetaxel	Breast	NCT02027376
Sonidegib	Bortezomib	MM	NCT02254551
Sonidegib	Gemcitabine+Nab-paclitaxel	Pancreatic	NCT01431794
Sonidegib	Paclitaxel	Ovarian	NCT02195973
Saridegib	Gemcitabine	Pancreatic	NCT01130142
Glasdegib	Cytarabine+daunorubicin+decitabine	AML	NCT01546038
Glasdegib	Azacitidine	AML	NCT02367456
BMS-833923	Dasatinib	CML	NCT01357655
BMS-833923	Cisplatin+capecitabine	Gastric/Esophageal	NCT00909402
BMS-833923	Carboplatin+etoposide	SCLC	NCT00927875
BMS-833923	Dasatinib	CML	NCT01218477
Taladegib	Paclitaxel+carboplatin+radiation	Esophageal	NCT02530437

Abbreviations: AML, acute myeloid leukemia; EGFR, epidermal growth factor receptor; SCLC, small cell lung cancer; BCNS, basal cell Nevus syndrome; MM, multiple myeloma; CML, chronic myeloid leukemia

Table 4.

Clinical trials examining smoothened antagonists combined with investigational compounds

Smo Antagonist	Targeted Therapy (Target)	Cancer	Identifier
Vismodegib	RO4929097 (gamma-secretase/Notch)	Breast	NCT01071564
Vismodegib	RO4929097 (gamma-secretase/Notch)	Sarcoma	NCT01154452
Vismodegib	GSK2256098 (FAK/focal adhesional kinase)	Meningioma	NCT02523014
Sonidegib	Buparlisib (PI3K)	BCC	NCT02303041
Sonidegib	Buparlisib (PI3K)	Solid tumors	NCT01576666

Abbreviations: BCC, basal cell carcinoma

Clinical researchers are employing the knowledge gained from pre-clinical studies and combining HH inhibition with new molecules targeting multiple signaling pathways and targets (Tables 3 and 4). Examples include the use of nivolumab, a PD-1 inhibitor, which is being combined with vismogedib in patients with basal cell/Nevus Syndrome (BCNS).⁹⁰ A novel FAK inhibitor, GSK2256098 was combined with vismogedib in patients with meningioma.⁹¹ Patients with advanced or metastatic BCC, who are either naïve to or refractory after SMO antagonist monotherapy, are being treated with sonidegib and buparlisib, a phosphoinositide 3-kinase inhibitor.⁹² Children and young adults with recurrent brain tumors are first having their brain tumors molecularly characterized, and then being stratified for combination therapies that include using sonidegib with ribociclib, a cyclin D1/CDK4 and CDK6 inhibitor.⁸² In a Phase II trial, ribavirin (a broad-spectrum anti-viral) and vismogedib are combined with and without decitabine in acute myeloid leukemia patients.⁹³ The SMO inhibitor, BMS-833923, is currently being combined with cisplatin and capecitabine for inoperable, metastatic gastric adenocarcinomas,⁹⁴ as well as carboplatin and etoposide in small cell lung cancer.⁹⁵ A triple cocktail of azacitidine, decitabine, combined with sonidegib is currently being evaluated in patients with myeloid malignancies.⁹⁶ While a monotherapeutic approach using SMO antagonists may not have been the most efficacious, combination therapy with chemotherapies (Table 3) or other investigational targeted therapies (Table 4), could provide an avenue for the continued and expanded use of HH inhibitors for other cancers, while circumventing refractory disease and drug resistance.

In addition to SMO antagonists, there is a growing body of pre-clinical research devoted to direct inhibitors of the downstream effectors of HH signaling, the Gli family of transcription factors. It is now understood that certain cellular pathways (e.g. PI3K, MAPK, Wnt, NF-κB, and TGFβ) can activate Gli transcription independently of the “classical” HH signaling pathway, also referred to as non-canonical signaling.^97,98 Compounds such as GANT58, GANT61, and Glabrescione B, which can directly inhibit the binding of Gli1 and Gli2 to DNA, have been found to be effective against cancer cells both in vitro and in vivo.^99,100 In an experiment performed by our lab, it was found that knockdown of both GLI1 and GLI2 using siRNA decreased ovarian cancer cell growth in vitro more than either knockdown alone (Figure 2). GANT61 (a Gli-1/Gli-2 antagonist) inhibited growth of ovarian cancer cells and decreased the expression of HH target genes (GLI1, GLI2, PTCH1) compared to vismodegib and sonidegib (Figure 3). Additional in vitro and in vivo studies have shown effectiveness of GANT61 in models of breast, lungs, and many other cancers.^101–104 Collectively, these studies suggest that direct inhibitors of Gli would be useful in cases where downstream activation of Gli transcription occurs independently of upstream PTCH repression and/or SMO activation, scenarios in which SMO antagonists would potentially be no longer therapeutic.

Figure 2.

Combined knockdown of GLI1 and GLI2 decreases ovarian cancer cell growth more than individual knockdown alone in vitro. Platinum-resistant A2780cp20 ovarian cancer cells were grown for up to 6 days after transfection with control siRNA, GLI1 siRNA, GLI2 siRNA or combined GLI1+ GLI2 siRNAs. Cell viability was determined using an MTT assay.⁶⁷

Figure 3.

Inhibition of the HH signaling pathway decreases cell viability of platinum-resistant A2780cp20 ovarian cancer cells. (a) Cell viability after 96 hours of varying doses of HH inhibitor (SMO antagonists, cyclopamine, vismodegib, sonidegib; GLI antagonist, GANT61) treatment using an MTT assay. Data are representative of 3 independent experiments. (b) Gene expression of GLI1, GLI2, and PTCH1 was examined in A2780cp20 ovarian cancer cells treated with HH inhibitors for 24 hours compared to untreated controls. Date are representative of 3 independent experiments

Another possibility for therapeutic intervention of HH signaling is for tumors that demonstrate overexpression of the HH pathway, particularly HH ligands, but lack activating mutations in PTCH1, SMO, or other HH components. In these tumors, non-canonical HH signaling may be playing a role in tumorigenesis. Faria et al.²⁹ showed PTCH-dependent, SMO-independent signaling was capable of activating protein kinase A (PKA) and Wnt signaling, and regulating SHH-mediated endocytosis. In a mouse model of small cell lung cancer, SHH overexpression induced chromosomal instability and promoted nuclear translocation of Cyclin B1 in cancer cells; whereas overexpression of constitutively active SMO did not.¹⁰⁵ Canonical HH signaling has been shown to suppress Wnt signaling in colon cancer cells and normal colon cells; however, in colon cancer organoids enriched for cancer stem cells (CSC) there was an increase in active Wnt signaling and PTCH1 expression.²⁸ These samples also expressed HH ligands and SMO, but lacked nuclear GLI expression. Additional experiments demonstrated that Wnt was activated via SHH-PTCH-dependent signaling that was independent of SMO or GLI, and necessary for the maintenance of colon cancer CSCs. Current molecules available that block the interaction between SHH and PTCH include: robotnikinin, a small molecule SHH inhibitor; 5E1 an SHH antibody, and RU-SKI 43, a small molecule inhibitor of hedgehog acyltransferase that blocks formation of functional SHH.^106,107 Non-canonical HH signaling indicates another point in this pathway for potential development of cancer therapeutics.

The role of HH signaling plays in initiation and progression of human cancer make it an attractive target for cancer therapeutics. Many ongoing clinical trials highlight the importance of tracking molecular responses to drugs, whether in relation to patient response or acquired resistance. Continued inclusion of molecular and genetic analyses in clinical trials will be essential in determining response and efficacy of targeted therapies, as well as aiding in the design in the next generation of HH inhibitors.

Funding Statement

This work was supported by the NIDCR under Grants [K99/R00-DE023826; NIDCRT90-DE022736-01(DART)].

Disclosure of interest

The authors report no conflicts of interest.