In the last several years, inhibition of the hedgehog signaling pathway (HHSP) for the treatment of locally advanced basal cell carcinoma (BCC) and metastatic BCC (mBCC) has led to a potential revolutionary change in how advanced BCC is treated. Patients with unresectable and/or mBCC are now personally seeking or being referred for clinical trial evaluation involving inhibitors of smoothened (SMO). Whereas BCC of the skin is the most common malignancy with over 1.4 million cases per year in the United States alone,1-3 mBCC is rare with an estimated incidence of <0.003% to 0.5%.4
Consideration of mBCC often is not included in the typical workup of a patient with multifocal metastatic disease who has a history of BCC. It is our hope that the current review, which includes evaluation of the timing from first BCC to metastasis, disease location, and radiographic features of the patients, will contribute to general knowledge about mBCC and awareness among providers, patients, and support groups along with providing an outlook for the future of treating mBCC. This review focuses on the clinical and radiographic findings in 22 patients with mBCC who were evaluated at an oncology clinical trials center after the advent of clinical trials with a SMO antagonist, along with a discussion of BCC epidemiology and pathogenesis, the HHSP, and current and future treatments for this rare presentation of the most common malignancy.
Basal Cell Carcinoma Overview
BCC is the most common malignancy and is a skin cancer. It is believed that BCC originates from the basal layer of the epidermis, the interfollicular epidermis, and the hair follicle. By using mouse models, investigators demonstrated that multiple epithelial compartments in skin can form BCC-like tumors, and the subtype of BCC can be manipulated, depending on which epithelial compartment is altered.5 It is estimated that approximately 65% to 75% of nonmelanoma skin cancer treated in the United States is BCC1, 2; and, in 2006, 2.2 million nonmelanoma skin cancers were treated.3
Approximately 95% of individuals are diagnosed with BCC between ages 40 and 79 years.6 BCC is much more common in Caucasians than individuals of African descent or dark-skinned populations.6 The incidence is approximately 30% higher in men than in women.7
Close to 90% of BCCs occur in the head and neck region, and risk factors include fair skin pigmentation, radiation (ultraviolet and/or ionizing), exposure to arsenic or polycyclic aromatic hydrocarbons, immunosuppression, scars, and certain genetic syndromes, such as nevoid BCC syndrome (NBCCS).8 A previous personal history of BCC leads to an increased risk of developing BCC in the future. BCC rates also are higher in individuals who live closer to the equator,7, 9 in the United States, and in other countries, such a Australia,10 in contrast to Northern European countries like Finland, which have a relatively lower incidence of BCC.11
Nevoid Basal Cell Carcinoma Syndrome
NBCCS is a rare, autosomal-dominant, multisystem disorder with a prevalence of 1 in 57,000 to 256,000 individuals. It results from germline mutations in the Patched gene (PTCH1) located on chromosome 9q22.3.12, 13 Not all patients with NBCCS have detectable PTCH1 mutations, suggesting the presence of mutations in other components of the HHSP.14 NBCCS is more common in Caucasians but has been reported in individuals of African12 and Asian decent.15 Characterized by a wide range of developmental abnormalities and a predisposition to neoplasms, the main clinical manifestations of NBCCS include early onset of multiple BCCs (median age, 20 years),12 odontogenic keratocysts of the mandible, palmar and plantar pitting, craniofacial anomalies, and skeletal abnormalities.13 Approximately 5% to 10% of individuals with NBCCS may develop medulloblastoma.13 An individual with NBCCS may have hundreds to thousands of BCCs treated, yet their life expectancy does not differ much from that of individuals without NBCCS.13
A diagnosis of NBCCS requires satisfactory clinical evidence of either 2 major criteria or 1 major and 2 minor criteria.16 The radiographic diagnostic workup for NBCCS may include the following: a skull x-ray or computed tomography (CT) scan to identify calcified falx cerebri16; panoramic films to identify odontogenic keratocysts17; and chest, hand, and/or foot x-rays to identify other skeletal abnormalities.16
Many histologic variants of BCC have been characterized. Two main categories exist that subdivide BCC by their aggressiveness. BCCs with nonaggressive growth patterns include nodular, superficial, and adnexal variants; whereas BCC with aggressive growth patterns include metatypical/basosquamous, morpheaform, infiltrating, sclerosing, and micronodular variants.6, 18, 19 It is noteworthy that, in mouse models, postnatal induction of the glioblastoma protein 2 (GLI2) activator in telogen follicle stem cells leads to the development of nodular BCC-like skin tumors.5 However, when the GLI2 activator is induced in the interfollicular epidermis or epidermis from hairless skin, superficial BCC-like tumors arise.5 In contrast to superficial BCCs, the hair follicle is a potential site of origin for nodular BCC in mice.5 In addition, the level of GLI2 activator can drive the development of nodular BCC-like tumors.5
Human BCC may also contain an admixture of histologies that fit into both growth pattern categories. Nearly all mixed category BCC originate in the head and neck region.18 Because there are mixed category BCC in humans and hair follicle stem cell progeny migrate into the epidermis during wound healing, murine models suggest an explanation of this with the possibility of nodular BCC tumors expanding upward to involve the epidermis as a superficial BCC.5 Detailed description and images of distinct features of BCC variants are beyond the scope of this review.6, 8, 19
In recurrent mixed category BCCs, it is the aggressive growth pattern variant that appears to predominate the patient's disease course.18 The incidence of mixed category BCC can range from 11% to 43%. In >85% of these tumors, the aggressive growth pattern features may be unsuspected.18 In other words, it is believed that the patient has a BCC variant with only nonaggressive growth pattern features based on the initial histopathologic assessment. This can lead to inadequate initial therapy, resulting in recurrent BCC and causing additional morbidity. Of the multiple modalities that can be used to treat primary BCC, each possesses differing rates of success (no BCC recurrence). Mohs' micrographic surgery has the highest rate of success, with a recurrence rate up to 3.3% rate.18
Metastatic Basal Cell Carcinoma
mBCC is exceedingly rare: The most conservative estimated rate is approximately 0.0028% or 28 cases per 1,000,000 BCC diagnoses, but the incidence has been cited as high as 0.5%.4 Thus, annually, the most conservative estimated rate of mBCC is approximately 40 cases per year in the United States.2, 3, 20 Nearly all reported cases of mBCC have been in Caucasians, and few individuals of African or Asian decent diagnosed with mBCC have been reported in the literature.21-23
In 1951, diagnostic criteria for mBCC were defined by Lattes and Kessler24 (Table 1). In mBCC, squamous features/differentiation appear to be more frequent than in primary BCC. However, as long as there are elements of BCC in the pathologic specimen, the diagnosis should remain classified as mBCC rather than metastatic squamous cell carcinoma.25 mBCC can spread through subcutaneous infiltration, hematologically, or, in approximately 70% of cases, through the lymphatic system.4, 27 Nearly all histologic variants of BCC have been reported in mBCC based on recent data from mixed category BCC18 and >80% of mBCC originate from primary BCC of the head and neck.25 Thus, we wonder whether nonaggressive variants may transform to aggressive histologic variants during disease progression. We suspect that the undetected aggressive growth pattern in most mBCCs may explain why nonaggressive histologies, such as superficial or nodular types, have been associated with mBCC. Future research may clarify this issue.
Table 1. Diagnostic Criteria for Metastatic Basal Cell Carcinoma
To be considered mBCC, all 3 criteria must be met.
1. Primary tumor originated in the epidermis or follicular skin
2. Tumor spread to a distant site, not local extension
3. Both primary and metastatic tumors have the histologic appearance of BCC but not solely squamous cell histology
Features of the Metastatic Basal Cell Carcinoma Patient Cohort
In total, 22 individuals with mBCC were evaluated (Tables 2 and 3). The majority of these patients traveled a distance even as much as >3300 miles (1 way). Of the 22 patients, there were 18 were men, and 10 patients had other family members with BCCs. Five of these individuals had confirmed NBCCS. The median age at first BCC diagnoses was approximately 43 years (range, 14-68 years). The majority of patients had primary BCC originating in the head and neck region.
Table 2. Detailed Clinical Features of the Current Metastatic Basal Cell Carcinoma Cohort
Age at First BCC, y
Histology Before Metastasis
Time to Metastasis, y
Prior Surgery/ XRT
Age at VGPCC Consultation, y
Symptoms at VGPCC Consultation
Best Response to Hedgehog Inhibitor
Abbreviations: 5FU, 5-fluorouracil; A, pemetrexed; B, bevacizumab; BCC, basal cell carcinoma; BCC, basal cell carcinoma; C, carboplatin; Cape, capecitabine; Cis, cisplatin; D, docetaxel; EGFR, epidermal growth factor receptor; EPHB4, ephrin type-B receptor 4; G, gemcitabine; HER2, human epidermal growth factor receptor 2; HSP90, heat shock protein 90; NA, not available or applicable; Nab-paclitaxel, nanoparticle albumin-bound paclitaxel; NBCCS, nevoid basal cell carcinoma syndrome; NOS, not otherwise specified; P, paclitaxel; PD, progressive disease; PI3K, phosphoinositide 3-kinase; PR, partial response; RT, radiotherapy; SD, stable disease; TRP, tumor-related pain; VEGF, vascular endothelial growth factor; VGPCC, Virginia G. Piper Cancer Center; XRT, external-beam radiotherapy; y, years.
This was the histologic subtype of all biopsied BCCs immediately preceding metastasis.
Metastases may have been present at initial diagnosis; however, complete staging imaging was not done until the patient was evaluated at an oncology clinical trials center.
Of 14 patients with primary BCC who had pathology reports that could be classified by variant, 1 patient had nonaggressive growth pattern features (nodular), 6 patients fulfilled the mixed category classification, and 7 patients had aggressive growth pattern features. The median time to developing mBCC was approximately 9 years (range, 1-48 years). The median age of the patients with mBCC was approximately 56 years (range, 46-89 years). Only 5 of 22 patients had no tumor or previous treatment-related symptoms at the time of their initial consultation. All but 1 patient had received previous BCC-related therapy, including surgery, radiotherapy, and/or chemotherapy. The sites of frequent mBCC involvement included lymph nodes, lung, skin, and bone. Other less common involved sites included pleura, liver, and brain. These results are similar to those reported in case series reviews (see Table 3) with a few exceptions. Our cohort had a higher prevalence of men with mBCC and a lower percentage with primary BCC involving the head/face; however, this may be because of our relatively smaller sample size. Finally, our cohort had a higher prevalence of lymph node and lung involvement, possibly because of improved imaging modalities compared with series reports preceding the 1990s. Imaging modalities like CT and 18F-fluorodeoxyglucose (FDG)-positron emission tomography (PET)/CT scanning now allow for more accurate assessment of disease extent of mBCC than ever before.
Radiographic Findings of Metastatic Basal Cell Carcinoma
All CT imaging features were assessed before patients received SMO antagonist therapy. Most patients with mBCC who were evaluated by CT scans had multiple sites of disease (Fig. 1a-f). Only a limited number of organ systems had metastases identified, as noted in Table 4. BCC metastases had variable appearances on CT imaging. Individual metastatic lesions ranged in size from a few millimeters to several centimeters. The CT densities of the tumor lesions generally were solid, but necrosis and cystic changes were not uncommon.
Table 4. Metastatic Basal Cell Carcinoma: Radiographic Sites of Involvement
No. of Patients, N = 22
Radiographic evidence of skin involvement was observed either as a single site or as multifocal sites with concurrent basal cell carcinoma involvement of a metastatic site.
Most metastatic disease either was centered within the skin and surrounding soft tissue and lymphatic system or spread hematogenously to the lungs, pleura, or bone. It is noteworthy, however, that the frequency of liver involvement was relatively low (1 patient had multiple liver metastases that measured approximately 1 cm in greatest dimension, and the other patient had small-volume serosal liver metastases) compared with lung and bone involvement. The attributes of mBCC that potentiated differential spread by lymphatics in 1 patient or through blood circulation in another have yet to be elucidated.
Skin and soft tissue involvement fell into 3 pattern types on CT imaging, including 1) an infiltrative pattern with loss of adjacent soft tissue planes (N = 5), 2) a nodular pattern with relatively well preserved tumor margins (N = 3), or 3) a mixed pattern (N = 1). The necrosis within lesions sometimes was observed, particularly in tumors with a nodular pattern.
Pulmonary disease on CT scans ranged from multiple lesions (1-3 metastases) to widespread parenchymal disease. The lesions tended to form coalescent masses in the lungs. Most lesions had spiculated margins, especially when they were large, whereas the smaller lesions (<1 cm in greatest dimension) tended to be more rounded and well defined. Central necrosis often was observed within the larger lesions, but cavitation was rare. All pulmonary segments were involved without a propensity for the right or left lung. Volume loss occasionally was observed in the setting of conglomerate lesions, suggesting a component of fibrosis and retraction. Pleural disease was noted in 18% of patients, usually in the form of moderate-to-large effusions. Nodular pleural disease was noted in 1 patient.
Bone lesions were identified in 8 of 22 patients (36%). Their appearance on CT scans varied from subtle osteolytic lesions (observed only on retrospect with PET/CT imaging in 1 patient), to mixed osteolytic and osteoblastic changes (similar to those noted in more common tumors, such as breast cancer), to destructive processes with a soft tissue component in 4 patients.
Lymph node involvement included predominately the neck (N = 6), axilla (N = 5), and chest lymph node stations (N = 8). Multiple lymph node station involvement was common (10 of 14 patients). The lymph nodes tended to have infiltrating margins and/or central low density on Hounsfield CT measurements, suggesting necrosis. Hyperenhancement was rare on contrast-enhanced CT images (N = 1).
All PET/CT imaging features were assessed before patients received SMO antagonist therapy. Very little is known about the FDG metabolic behavior of BCC metastasis. In theory, PET/CT images should be able to provide information that is complementary to that provided by conventional imaging techniques because of the functional nature of PET/CT scanning and whole-body surveys for disease. Hypermetabolic primary BCC originating in the head and neck region has previously been reported.28
In our cohort, we noted that patients who underwent PET/CT imaging had hypermetabolic activity in BCC lesions (Fig. 2a-e). Indeed, at least 1 hypermetabolic lesion was observed in every patient who had a baseline PET/CT scan before the initiation of SMO antagonist therapy. The peak FDG activity, measured as the maximal standardized uptake value (SUVmax), ranged from 1.9 to 16.8 (mean SUVmax, 7.3). Typically, this value was 2 to 4 times above background tissue metabolism. The distribution of FDG-avid disease within organ systems was nearly identical to that observed on contrast-enhanced CT images in those patients who underwent both types of imaging examinations. However, hepatic disease was not observed on PET/CT images in the 1 patient who had serosal disease on a CT scan. This may be because of the relatively small size and serosal nature of this patient's hepatic disease. PET/CT images tended to demonstrate more disease in bone and soft tissues compared with contrast-enhanced CT images, likely because of whole-body coverage with PET/CT scanning and the often subtle changes in the osseous structures involved in bone metastasis. Contrast-enhanced CT images tended to demonstrate more disease in the lung, especially in lesions that measured <7 mm. Failure to detect very small lung parenchymal lesions is a known limitation of PET/CT imaging. Nevertheless, PET/CT screening for patients with mBCC may be helpful to fully evaluate disease burden and to potentially measure therapeutic response.
Staging Workup and Treatment
The tumor, lymph node, metastasis (TNM) staging system is used commonly for most malignancies. A TNM staging system endorsed by the American Joint Committee on Cancer (AJCC) is available.29 When possible, for locally advanced BCC and mBCC, the TNM staging may enable better prognostication as newer systemic treatment modalities become commercially available and as awareness about advanced BCC increases.
Molecular Pathogenesis of Metastatic Basal Cell Carcinoma
Similar to primary BCC and NBCCS, the main genetic aberration in mBCC is most commonly the result of mutations or loss of heterozygosity (LOH) of the tumor suppressor PTCH1.30 In addition, mutations in smoothened (SMO) or mutations or LOH of PTCH2 also can lead to basal cell carcinogenesis.31-33 Together, these perturbations account for up to 95% of basal cell carcinogenesis.32, 33 In addition, TP53 gene mutations are identified in >50% of BCCs, although these probably are related to disease progression rather than carcinogenesis.32 PTCH1, PTCH2, and SMO are all components of the HHSP, which has several therapeutics currently in clinical development.
The Hedgehog Signaling Pathway
Targeting of the HHSP as an anticancer therapy is a fairly recent concept. The impetus that led to the HHSP discovery and characterization began with the initial observation of severe holoprosencenphaly and other congenital malformations in newborn lambs after pregnant ewes consumed California corn lily (Veratrum californicum) in 1962.34 The cause of these teratogenic effects was a chemical aptly named cyclopamine.35 Whereas normal HHSP signaling is involved in cell growth and congenital development,36, 37 is has been proposed that dysregulated HHSP signaling involves nearly 25% of malignancies.36 Ensuing research led to the demonstration that cyclopamine or synthetic HHSP pathway inhibitors potentially could be useful to inhibit HHSP signaling in malignancy.38
The main ligand that activates the HHSP pathway is Sonic.33, 39, 40 Signaling is initiated when Sonic binds to its receptor, PTCH1, and inactivates it.33, 39, 40 A series of events ensue, including disinhibition of the transmembrane domain protein SMO. SMO migrates to primary cilia and enables GLI activator,41 leading to transcriptional up-regulation of target genes.42 Without HHSP ligand signaling, PTCH1 remains activated, inhibiting SMO, leading to GLI repressor, such that HHSP target gene transcription is blocked (for further details, see Weiss and Von Hoff, 201033).
In cancers with PTCH1-activating and/or SMO-activating mutations, HHSP signaling is ligand-independent (Type I system) and, thus, does not require Sonic binding to PTCH1. Tumors comprising the Type I system include BCC, medulloblastomas, and rare variants of rhabdomyosarcoma.33, 40 The HHSP ligand-dependent autocrine cancers (Type II system) include breast, pancreas, lung, prostate, upper gastrointestinal, melanoma, and colorectal cancers.33, 40 The HHSP ligand-dependent paracrine cancers (Type III system) include colorectal, pancreas, and prostate cancers.33, 40
For the most part, in the pre-HHSP inhibitor era, treatment largely has been disappointing, with a median survival of 8 months (range, 0-192 months).25 By using cisplatin-based therapy, some individuals have enjoyed prolonged disease-free survival.43-46
Targeting the Hedghog Signaling Pathway With Smoothed Antagonists
Targeting the HHSP in cancer therapeutics is an active field in drug development. Recently, vismodegib (GDC-0449) was approved by the US Food and Drug Administration for treating BCC that has recurred after surgery or that has advanced locally or metastasized. There are now at least 7 additional SMO antagonists in various stages of clinical trial development, including BMS-833923, IPI-926, LDE-225, PF-04449913, LEQ506, TAK-441, and LY2940680. The most impressive and positive, life-altering results from these therapeutic trials in patients with metastatic cancer have been document in mBCC. At least 2 SMO antagonists have undergone phase 2 evaluation in advanced BCC trials.47
The first human results in advanced/mBCC demonstrating a response to vismodegib were published in 2009.30 Overall, this agent is well tolerated with primarily mild-to-moderate side effects, including dysguesia, hair loss, and muscle cramps.48 Subsequently, the single-arm phase 2 results of vismodegib in mBCC (N = 33) recently were announced, and the agent demonstrated a 30% response rate and a median progression-free survival of 9.5 months.49 Results for LDE225, the other SMO antagonist in phase 2 trials for advanced BCC, are still too early to report. However, in the phase 1 study, 5 patients with advanced BCC attained at least a partial response, and at least 2 patients were on therapy for >4 months.50
In total, 20 patients with mBCC in this cohort (Table 2) received treatment with SMO antagonists. The best response achieved was a partial response in 8 patients (for a representative example, see Fig. 3a,b), stable disease in 10 patients, and 2 patients had progressive disease as their best response to an SMO antagonist.
Other Drugs Targeting the Hedgehog Signaling Pathway
Other naturally occurring agents or drugs that have been approved for the treatment other diseases have demonstrated the ability to inhibit HHSP inhibition in preclinical studies.51 Vitamin D3 (cholecalciferol) is a fat-soluble vitamin that can inhibit SMO and has greater potency than cyclopamine in vitro.52 In preclinical models, drugs that inhibit GLI1 or forkhead box protein M1 (FOXM1), which are downstream of SMO, also have been reported. By targeting downstream of SMO, in theory, mBCC tumors that lack sensitizing genetic aberrations that would be responsive to SMO antagonists or mBCC tumors that develop resistance to SMO antagonists may derive benefit from these agents. Agents that target GLI1 include GLI antagonist 61 (GANT61)53 and arsenic trioxide.54 AY9944, a diamine substructure of GANT61 and an inhibitor of the enzymatic activity and transcriptional inducer of 7-dehydrocholesterol reductase (Dhcr7), can attenuate HHSP signaling when there is up-regulation of endogenous Dhcr7.55 Thus, other Dhcr7 modulators, such as the antidepressant imipramine and antipsychotics like clozapine, chlorpromazine, and haloperidol; can regulate HHSP signaling.55 Itraconazole, an antifungal agent, also inhibited GLI in a reporter assay.56 In fact, a trial of oral or topical itraconazole in nonmetastatic BCC is underway (National Clinical Trial NCT01108094). Agents that target FOXM1 include thiazole antibiotics like siomycin A and proteosome inhibitors like bortezomib.51 Statins and human 3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors, which are approved to prevent and treat heart disease by lowering blood cholesterol, also were able to reduce HHSP target gene transcription and pathway-dependent proliferation in medulloblastoma in vitro.57
In conclusion, during the last 5 years, there has been a plethora of new agents targeting SMO, a key component of the HHSP. For the majority of mBCCs, SMO antagonists should be a great context in which to match a targeted agent to tumor genetic vulnerability. The emergence of disease progression in mBCC caused either by selection for resistant tumor clones or by the development of acquired resistance is a real clinical problem for patients with mBCC.33, 57 Like in other targeted therapies for uncommon malignancies, such as chronic myelogenous leukemia and gastrointestinal stromal tumors, it is likely that next-generation HHSP inhibitors will be introduced to combat mBCC in patients who are nonresponsive or who progress on current SMO antagonists. With improved awareness of mBCC, we hope that, at the least, patients who have this very rare cancer are considered for referral to a cancer center that is investigating SMO antagonists in mBCC or in advanced solid tumors.
We thank the patients and clinic staff, including Daniel Von Hoff, Mitesh Borad, Raoul Tibes, Ramesh Ramanathan, Lisa Blaydorn, Molly Downhour, Katy Schroeder, Lisa Rumble, Karen Lewandowski, Barry Kruse, and Valerie George; and we thank Maya Hernandez and Curtis Thacker for their assistance with assembling scans for review.
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURES
G.J.W. is a principal investigator or coinvestigator on trials involving hedgehog pathway inhibitors in cancer with Genentech, Infinity Pharmaceuticals, Novartis, and Eli Lilly. These companies provide funding to his institution. G.J.W. is also on the Speaker's Bureaus of Eli Lilly and Genentech.