Overcoming crizotinib resistance in ALK-rearranged NSCLC with the second-generation ALK-inhibitor ceritinib

ABSTRACT

In up to 5% of non-small cell lung cancer (NSCLC) patients, the EML4-ALK translocation drives tumor progression. Treatment with the ALK inhibitor crizotinib is more effective than standard chemotherapy. However, resistance to crizotinib occurs after approximately 8 months. Ceritinib is the first second-generation ALK inhibitor approved for treatment of crizotinib-resistant NSCLC. Ceritinib inhibits two of the most common ALK-mutants that confer resistance to crizotinib: L1196 M and G1269A. Cells with ALK expression are more sensitive to ceritinib than crizotinib (IC₅₀ 25 nM vs. 150 nM, respectively). Alternative second-generation ALK inhibitors such as Alectinib, Brigatinib and PF-06463922 are currently in development, each affecting different crizotinib-resistant ALK target mutations. Genetic identification of crizotinib-resistant mutants is essential for selecting the optimal treatment strategy in NSCLC patients to overcome resistance and to increase progression-free survival.

KEYWORDS:

• Acquired resistance
• tyrosine kinase inhibitors
• ceritinib
• crizotinib
• anaplastic lymphoma kinase
• resistance mutation. EML-ALK

Introduction

In 2015, 220,000 new lung cancer patients are predicted in United States of America, of which 85% would consist of non-small cell lung cancer (NSCLC).[1] In up to 5% of these NSCLC tumors, the fusion-type protein kinase EML4 (echinoderm microtubule-associated protein-like 4)-ALK (anaplastic lymphoma kinase) is the main oncogenic driver. This ALK-rearrangement is generated as a result of a small inversion within the short arm of human chromosome 2.1.[2]

The EML4-ALK fusion oncogene mediates ligand-independent dimerization of ALK, resulting in constitutive protein kinase activity. Cell culture systems with EML4-ALK possess potent oncogenic activity.[2] ALK belongs to the insulin receptor (InsR) superfamily and is similar to leukocyte tyrosine kinase (LTK). Under normal conditions, ligand binding results in dimerization of ALK, trans-phosphorylation and protein kinase activation. Fusion of EML4 with ALK enables ALK to dimerize ligand independently, resulting in an aberrant and constitutively active protein. Aberrant expression of the EML4-ALK fusion protein activates several canonical pathways such as the PI3K pathway, the JAK3/STAT pathway and the Ras/Mek/Erk pathway. These pathways are important for cell survival and cell proliferation, respectively (Figure 1).[3] Inhibition of EML4-ALK leads to downregulation of Ras/Mek/Erk and PI3K/Akt and subsequently cell death, indicating the importance of the activation of these pathways in EML4-ALK fusion-bearing cancers.[4] In transgenic mouse models, lung-specific expression of EML4-ALK leads to the development of numerous lung adenocarcinomas. Treatment with a specific ALK-inhibitor strongly reduced the number of nodules in mouse lungs.[5] These results demonstrate the importance of the EML4-ALK fusion kinase in the carcinogenesis of NSCLC that possess this fusion kinase.

Figure 1. EML4-ALK fusion oncogene and downstream signaling pathways. Information from Shaw et al. [3] and Roskosky.[6]

Current treatment

The identification of PF-02341066, or crizotinib (Xalkori™), as an effective ALK-inhibitor led to its use in EML4-ALK-positive NSCLC patients. Kwak et al. [7] reported a 57% overall response rate in these patients after oral administration of crizotinib. In the PROFILE1014 study, crizotinib was superior in the treatment of patients with previously untreated, advanced NSCLC with ALK rearrangement in comparison to standard chemotherapy for NSCLC. Unfortunately, all patients eventually relapsed, after an average response duration of 11.3 months.[8] Interestingly, the central nervous system is one of the most common sites of relapse in ALK-positive crizotinib-resistant NSCLC patients due to the inability of crizotinib to target CNS lesions.[9,10] CNS relapses in crizotinib-treated patients are likely caused by poor penetration of the blood–brain barrier by crizotinib. This effect is supported by the low intracranial response rate of crizotinib in ALK-rearranged NSCLC patients (7% ORR).[11] In addition, crizotinib concentrations in the cerebrospinal fluid (CSF) of a 29-year-old NSCLC patient with EML4-ALK rearranged were remarkably low, exhibiting a ratio of the CSF concentration of crizotinib to the plasma concentration of only 0.0026.[12] Crizotinib is an excellent substrate for p-glycoprotein (P-gP). Therefore, the low penetrance of crizotinib in the CNS is not caused by a poor uptake but by an efficient efflux mediated by P-gP located in the capillary epithelial cells of the blood–brain barrier.[13]

Mutations in the ALK tyrosine kinase domain that confer resistance to crizotinib have been identified in approximately one third of the re-biopsied tumors of patients with ALK-rearranged NSCLC.[14] Katayama et al. identified four resistance mutations that eventually led to crizotinib resistance: three mis-sense mutations L1196M, G1202R, S1206Y, and an amino acid (threonine) insertion mutation (1151Tins). All four mutations conferred resistance to crizotinib.[15] More specifically, all mutations were located in the ATP-binding pocket of ALK, implicating that resistance to crizotinib is caused by disruption of the ALK-ATP-binding site. Thus, the identification of drugs that can inhibit the growth of recurrent tumors, specifically directed toward such mutations, is highly warranted.

Characteristics of ceritinib

Identification of an additional ALK-inhibitor was done by Novartis Pharmaceutical Corporation after optimization of their lead compound TAE684 [16] (Figure 2). Further modification of this molecule was performed by reversing the piperidine para position of the aniline moiety, replacing the methoxy moiety by an isopropoxy moiety and adding a methyl group at the position para to the isopropoxy moiety. This led to the development of the ALK-inhibitor LDK378 or Ceritinib (Zykadia™).[17]

Figure 2. Information for structural formulas of crizotinib, the lead compound TAE684 and ceritinib (LDK378) taken from Chen et al. (2013) [17]; other structures from references cited in the text.

Ceritinib showed a high potency against ALK enzymatic activity with an IC50 value of 200 pM. In addition, ceritinib only showed inhibition against IGF-1R, InsR, and STK22D out of a panel of 46 kinases with a minimum selectivity of 70-fold. In Ba/F3-NPM-ALK (ALK expressing cells) cells, ceritinib inhibited growth with an IC50 of 26 nM, which is 6-fold more potent than crizotinib (150.8 nM).[18]

There are several reasons that can explain the ability of ceritinib to overcome crizotinib resistance, in addition to the higher efficacy of ceritinib compared to crizotinib to inhibit the EML4-ALK fusion protein enzyme activity. In some patients, the low crizotinib plasma levels are not sufficient to inhibit oncogenic activity which is overcome by ceritinib. In addition, one of the gatekeeper mutations that causes crizotinib resistance (L1196M) can be inhibited by ceritinib. In contrast to crizotinib, ceritinib can interact with large and lipophilic gatekeeper mutants at this position, thereby inducing ALK inhibition. The likely reason is that the polar aromatic amine in the 2-position of the pyridine scaffold of crizotinib is not able to make interactions with a large lipophilic residue at the gatekeeper position while the chlorine in the 5-position of the pyrimidine unit of ceritinib may interact more favorably.[19]

Ceritinib is not able to overcome crizotinib resistance in tumors harboring the resistance mutation G1202R. The G1202R mutation leads to steric hindrance for both drugs due to a larger charged side chain (guanine to arginine). Ceritinib can only partially overcome resistance to crizotinib of tumors harboring C1156Y.[20]. One of these mutations is identified in 5 out of 11 biopsies from patients that acquire ceritinib resistance after treatment. However, ceritinib was particularly effective in overcoming crizotinib resistance in tumors harboring L1196M, G1269A, I1171T, and S1206Y mutations.[21] Also in BA/F3 models transfected with these mutants, ceritinib was unable to reverse resistance to crizotinib in the C1156Y, 1151Tins, L1152P, F1174C, and G1202R mutants, but was about 100–1000 times more potent in L1196M, S1206Y, G1269A, and G1269S mutants and somewhat less against I1171T mutants.[20]

By evaluating the difference between the molecular structures of ceritinib and crizotinib, it was possible to explain how ceritinib could overcome the most common mutations that lead to crizotinib resistance (Figure 3). The G1269 site is situated just proximal to D1270 of the activation loop DFG-motif. A mutation to alanine (Ala) in the common crizotinib mutant G1269A precludes binding of crizotinib since the phenyl ring of crizotinib does predict a steric hindrance. However, there is no steric hindrance for the binding of ceritinib. In addition, at L1196, the Cl moiety of ceritinib can interact hydrophobically with the leucine (Leu) side chain, while in the L1196M mutant, interaction of methionine (Met) and the Cl moiety of ceritinib is still possible. In contrast, crizotinib binding is adversely affected by the gatekeeper L1196M mutation due to steric interference. These structural implications support the increased potency of ceritinib in these crizotinib-resistant mutants.[21]

Figure 3. Structure of human ALK. The left structure (A) represents a dormant enzyme, with in the N-lobe the β-strands labelled at 1-5. Beneath the glycine rich-loop a space-filling model of staurosporine (a general protein kinase inhibitor) is shown, which occupies the ATP binding site. In the middle structure (B) the C- spine and R-spine denote the residues that constitute the catalytic and regulatory spines. The ALK gatekeeper Leu1196 contacts both spines. In this gate-keeper the Met side group of the L1196M mutation can still interact with the Cl moiety of ceritinib but not with the 2-amino group and the alkoxy moiety of crizotinib because of steric interference. The right structure (C) shows the interactions between the human ALK catalytic core residues, ATP and the protein substrate. The G1269M mutation is just proximal of D1270 in the activation DFG motif; steric hindrance of phenyl ring of crizotinib leads to resistance, but can still be inhibited by ceritinib. Reproduced with permission from Roskosky.[6]

Clinical trials with ceritinib

Inclusion criteria of the first Phase I clinical trial with ceritinib were stage IV NSCLC patients with presence of ALK translocations as demonstrated by FISH in at least 15% of tumor cells. The trial consisted of a dose-escalation phase, in which 59 patients received ceritinib at 50–750 mg orally once daily, and a dose expansion phase in which 130 patients with crizotinib resistance or intolerance were treated with ceritinib. In 80 patients who received crizotinib treatment prior to ceritinib, the response rate was 56%. Interestingly, patients with untreated CNS lesions after crizotinib treatment showed response to ceritinib.[20,22]

Ceritinib also showed similar efficacy in crizotinib-treated patients without ALK rearrangements, implying that crizotinib-resistant tumors may remain ALK-dependent which would be susceptible to a more potent ALK-inhibitor such as ceritinib.[22]

During the dose-escalation phase, dose-limiting adverse reactions were observed in six patients (diarrhea, nausea, vomiting, hypophosphatemia, and transaminitis). The 750 mg daily dose was chosen as a recommended dose due to the frequency of persistent adverse reactions when escalating the dose to 900 mg once daily. About 163 patients with metastatic ALK-positive NSCLC with crizotinib resistance or intolerance received 750 mg ceritinib once daily in the expansion phase. Interestingly, the response rate in patients treated with more than 400 mg daily was 58%, implying that the current recommended dose is not necessary and a lower dose might be equally effective. The objective response rate (ORR; complete and partial responses) was used as the primary efficacy end point and duration of response as a secondary efficacy end point.

Safety of ceritinib

Safety evaluation was based on 255 ALK-positive patients (246 with NSCLC and 9 patients with other cancers who received ceritinib at a dose of 750 mg daily). Data obtained in Japanese patients with cancer and healthy volunteers were used to augment the safety evaluation.

The median duration of exposure to ceritinib was 6 months. There were no absolute contraindications identified for ceritinib that would outweigh its potential benefits in the indicated population.

At the proposed recommended dose of 750 mg orally (taken once daily without food), 98% of patients experienced gastrointestinal (GI) adverse reactions, namely diarrhea (86%), nausea (80%), vomiting (60%), and abdominal pain (54%). However, in the FDA report, it is also mentioned that taken with food will improve drug exposure and will decrease severity of gastro-adverse reaction of ceritinib [22] Additional common adverse reactions were fatigue, decreased appetite, and constipation (<25% each). Patients receiving ceritinib also experienced increased alanine transaminase (ALT) (80%), increased aspartate transaminase (AST) (75%), increased creatinine (58%), increased glucose (51%), increased lipase (29%), and increased bilirubin (16%). Serious and sporadically fatal adverse reactions included hepatotoxicity, interstitial lung disease, prolongation of the corrected QT interval, and hyperglycemia (Table 1). Other serious adverse reactions include pancreatitis, which was identified due to clinically relevant elevations of serum lipase.

Table 1. Common (>10%) or >2% NCI CTCAE* grade 3–4 adverse reactions.

	Crizotinib		Ceritinib
Adverse drug reactions	All grades (%)	Grade 3–4 (%)	All grades (%)	Grade 3–4 (%)
Gastrointestinal disorders
Diarrhea	60	0	86	6
Nausea	55	1	80	4
Vomiting	47	1	60	4
Abdominal pain	8	0	54	2
Constipation	42	2	29	0
Esophageal disorder^a	8	0	16	1
Neurological disorders
Neuropathy	19	na	17	na
Eye disorders
Vision disorder	60	0	9	na
Metabolic disorders
Decreased appetite	27	na	34	1
General disorders
Fatigue^b	27	na	52	5
Skin and subcutaneous disorders
Rash^c	9	na	16	0
Respiratory disorders
Interstitial pneumonitis	26	0	4	3

*National Cancer Institute Common Terminology Criteria for Adverse Events.

^aEsophageal disorder (dyspepsia, gastroesophageal reflux disease, dysphagia).

^bFatigue (fatigue, asthenia)

^cRash (rash, maculopapular rash, acneiform dermatitis).

na: data not available.

Sixty percent of patients required at least one dose reduction after administration of the recommended dose of 750 mg once daily. The most frequent adverse reactions that led to dose reductions were elevated ALT (29%) and elevated AST (16%). Ten percent of patients permanently discontinued ceritinib due to adverse reactions, most commonly pneumonia/pneumonitis and decreased appetite. Currently, a dose-optimizing study is being performed to improve the dose regimen of ceritinib in NSCLC patients with crizotinib-resistant EML4-ALK translocations.[20]

Several distinct differences can be seen when comparing the safety and toxicity profiles of ceritinib and crizotinib. The ceritinib-treated group had a higher percentage of patients presenting with all grade GI disorders with the exception of constipation (29% in ceritinib, 42% in crizotinib), higher percentage of patients with fatigue (52% vs. 27%), but a lower percentage of patients with vision disorders (9% vs. 60%) and respiratory disorders (4% vs. 26%). Interestingly, the percentage of patients with grade 3–4 interstitial pneumonitis was higher in the ceritinib-treated population (3%) compared to the crizotinib-treated patients (0%).

Pharmacokinetics

According to the FDA reports,[22] both crizotinib and ceritinib reach the maximum concentration (T_max) at 4–6 hours after a single dose and the steady state is reached at 15 days for both crizotinib and ceritinib (250 mg daily vs. 750 mg daily, respectively) (Table 2). In addition, the half-life of the two ALK-inhibitors after a single dose is also similar (at 250 mg crizotinib, 42 hours; at 750 mg ceritinib, 41 hours). Both ALK inhibitors have a lower clearance at steady state (crizotinib 60 L/h, ceritinib (33.2 L/h) compared to a single dose (crizotinib 100 L/h, ceritinib (88.5 L/h), which could be contributed to autoinhibition of CYP3A after multiple dosing. Plasma protein binding is 91% for crizotinib and 97% for ceritinib, both independent of drug concentration. Ceritinib also displays a preference for red blood cells over plasma (blood-to-plasma ratio of 1.31), in contrast to crizotinib which has a blood-to-plasma ratio of 1. However, preliminary results suggest that the blood-to-plasma ratio of crizotinib increases over time, leaning more toward red blood cells (paper in preparation). It is unknown whether the ceritinib blood ratio changes over time in a similar fashion. The exposure increased significantly when the drug is given with a high-fat meal with an increase in the area under the curve (AUC) of 73% and of the C_max by 41%. Also with a low-fat meal an increase in the ceritinib AUC of 58% and of the C_max 43% was found after a single 500-mg dose administered to healthy subjects.[22] This underlines the importance of food in administration of tyrosine kinase inhibitors.

Table 2. Pharmacokinetics.

	Crizotinib (250 mg daily)	Ceritinib (750 mg daily)
Time to maximum concentration (Tmax)	4–6 hours	4–6 hours
Time to steady state (Tss)	15 days	15 days
Steady state concentration (ng/mL; µM)	100 – 135 ng/ml	800 ng/ml
AUCinf (ng/mL/h)	2192 – 2946 ng.h/ml	-
AUC0-24 h (ng/mL/h)	-	16500 ng/ml/h
Half-life (T1/2)	42 h	41 h
Clearance (Cl) single dose Steady state	100 L/h	88.5 L/h
	60 L/h	33.2 L/h
Bioavailability (F)	43%*	NR
Plasma protein binding	91%	97%
Volume of distribution after single dose (Vd/F)	1772 L (50 mg I.V.)	4230 L (750 mg orally)
Mean accumulation ratio	4.5	6.2
Excretion (unchanged)	53%feces/2.3% urine	68%feces/1.3%urine
Metabolization	CYP3A4/5	CYP3A

*Calculated by comparing I.V. and oral administration with the assumption that hepatic clearance was identical [23]

The solubility of both crizotinib and ceritinib is poor as pH is increased making it difficult to determine the exact bioavailability. It was reported that crizotinib had an absolute bioavailability of 43% after comparing both 50 mg I.V. administration and 250 mg oral administration with the assumption that hepatic clearance was identical.[23] No data has been reported on the bioavailability of ceritinib. Although the specific cytochrome-P (CYP) enzymes for ceritinib are unknown, both drugs are metabolized by CYP3A enzymes, with crizotinib primarily by CYP3A4/5. Both crizotinib and ceritinib are primarily excreted through the feces (53% vs. 68% unchanged drug, respectively), while 2.3% and 1.3% of the administered dose of crizotinib and ceritinib, respectively, and were found in urine.

The volume of distribution (V_d) is reported differently in the two FDA reports. For crizotinib, V_d was measured after a single intravenous dose of 50 mg, while the V_d for ceritinib was calculated after single (oral) dose of 750 mg. If we assume these numbers are correct, at least twice as much ceritinib is needed to reach the optimal concentration in plasma, considering ceritinib has a V_d twice as large as crizotinib. However, ceritinib is 6 times more potent in inhibiting the growth of Ba/F3 ALK-positive cell lines than crizotinib, implying that a 6-fold lower concentration is as effective in vitro in ALK-inhibition as crizotinib. Furthermore, ceritinib is cleared approximately twice as slow as crizotinib at steady state (33.2 L/h vs. 60 L/h, respectively), which means that a higher dose of crizotinib would be necessary in order to maintain an optimal concentration. One important missing factor is the bioavailability of ceritinib, for without it, it is not possible to predict the total exposure (AUC) of ceritinib and compare it to crizotinib in a standard dose regimen. However, considering the frequent dose reductions (60% of the patients at 750 mg daily), we can assume that a lower dose of ceritinib might still be effective in treating crizotinib-resistant NSCLC with less adverse effects. In addition, the response rate of patients treated with more than 400 mg ceritinib was reported to be 58%, further supporting the efficacy of a lower recommended dose of ceritinib.

Furthermore, preliminary results indicate that crizotinib is accumulated in the lysosomes, isolating it from its target kinase ALK.[24] Because of the similarity in chemical structure to crizotinib, it can be hypothesized that a similar reduction of exposure to ALK for ceritinib in NSCLC patients could occur. More importantly, this effect could adversely affect the accuracy of V_d levels as well as the AUC, preventing accurate PK modeling.

Next-generation ALK TKIs

The discovery of ALK inhibitory agents and their efficacy against ALK-positive NSCLC led to the discovery of several potent ALK inhibitors designed to have a higher efficacy than crizotinib. While most ALK inhibitors are in development, only ceritinib recently gained FDA approval.[20] One of the inhibitors in development is Alectinib (RO5424802/CH5424802), designed by Roche to be more selective and potent at ALK inhibition than crizotinib.[25,26] Alectinib also has a selective activity against LTK and cyclin-G-associated kinase. However, unlike ceritinib, it is not active against IGF-1R and INSR. In addition, the activating ALK mutations F1174L and R1275Q found in neuroblastoma are susceptible to alectinib, suggesting a potential use for alectinib in neuroblastoma patients [27] (Table 3). A recent clinical trial showed that alectinib has an ORR of 55% (24 of 44 patients) in crizotinib-resistant patients with doses ranging between 300 and 900 mg twice daily.[28] Also in 52% (11 of 21 patients) of the patients with a CNS metastasis, a response was observed. A continuation of this phase 1/2 study is currently ongoing to further determine the recommended dose. Furthermore, alectinib treatment suggests efficacy in decreasing CNS tumors in crizotinib-treated NSCLC patients similar to ceritinib.[26] Common toxicities of alectinib include dysgeusia, elevated aspartate aminotransferase, and increased bilirubin (Table 3).

Table 3. Selectivity and target mutations of next-generation ALK-inhibitors.

Drug	IC50 ALK in vitro NSCLC	Target kinases	Selective activity against crizotinib-resistant mutations	Reference
Ceritinib	25 nM	ALK, IGF-1R, INSR, STK22D	LL119M(+), G1269A(+), I1171T(+), S1206Y (+), G1202R(−), F1174C(−)	[22]
Alectinib	53 nM	ALK, LTK, GAK	L1196M(+), C1156Y(+), F1174L(+), R1275Q(+)	[27]
Brigatinib	21 nM	ALK, EGFR, ROS1	L1196M(+) (ALK), T790M(+), (EGFR)	[29,30]
PF-06463922	2.7 nM	ALK,ROS	L1196M, G1269A, 1151Tins, F1147L, C1156Y, L1152R, S12026Y	[31]

GAK = cyclin-G-associated kinase; LTK = leukocyte tyrosine kinase.

Brigatinib (or AP26113) is currently in development by Ariad©. It is a highly potent ALK inhibitor (IC50 of 21 nM in a Ba/F3 model of EML4-ALK) with additional selectivity for EGFR and ROS1 (Table 3).[29] Brigatinib is able to inhibit ALK signaling in the presence of the L1196M gatekeeper mutation but also the T790M gatekeeper mutation of EGFR, which can be found in 50% of EGFR-mutant NSCLC treated with EGFR TKI.[32] In a phase 2 trial containing 24 ALK-positive NSCLC patients, 62.5% had an objective response to treatment of brigatinib with doses between 60 mg and 240 mg daily, while 9%–12% of patients presented with respiratory symptoms on day 1 or 2 after a dose of 180 mg daily. The recommended dose is therefore 90 mg daily and, if no pulmonary symptoms arise, can be increased to 180 mg daily.[33]

PF-06463922, which is currently in development by Pfizer, is a potent inhibitor of both ALK and ROS1, and is active against all known clinically relevant ALK and ROS1 mutants in crizotinib-resistant preclinical models (Table 3). PF-06463922 treatment in mice with EML4-ALK-driven brain tumors showed regression and increased overall survival.[31,34] A recent study supported these findings, demonstrating that PF-06463922 has high potency across ALK variants in vitro, inhibits ALK more efficiently than crizotinib, and more importantly: PF-06463922 treatment of both crizotinib-resistant and sensitive xenograft mouse models of neuroblastoma induced complete tumor regression.[35] Furthermore, Mologni et al. established several anaplastic large-cell lymphoma cell lines (which also harbor ALK translocations) resistant to the novel second-line ALK inhibitor ASP3026. PF-06463922 inhibited the growth of any resistant clone while other ALK inhibitors (ceritinib, alectinib, crizotinib, and brigatinib) were unable to completely impair outgrowth.[36] In a Phase I clinical trial with 24 ALK or ROS-positive patients, 6 patients with progressive disease after crizotinib or ceritinib treatment showed a partial response to PF-06463922 treatment, of whom 5 patients showed intracranial response (Table 4).[37] These results suggest a role for PF-06463922 as an effective drug in patients with ALK-driven lung disease, including patients who relapse after clinically available ALK-inhibitor therapy and/or patients who exhibit CNS disease.

Table 4. Overview of clinical trials currently in progress for next-generation ALK inhibitors (>20 patients).

Drug	Trial design	Crizotinib-resistant, naïve, or both	ALK-Positive patients	PFS (months)	Objective response rate (%)	Common adverse effects	Source
Crizotinib	Phase 2	–	172	7.7	65	Vision disorder, diarrhea, nausea	[41]
Ceritinib	Phase 1/2	Both	114		58	Nausea, diarrhea, fatigue	[14,22]
		resistant	80	6.9	56
	Phase 2	Naïve	124	11.1	63.7	Nausea, diarrhea, vomiting	[42,43]
		resistant	140	5.7	38.6
Alectinib	Phase 1/2	Naïve	46	NR	93.5	Dysgeusia, AST, Bili	[26]
	Phase 1/2	Resistant	37	NR	59	NR	[40]
	Phase 1/2	Resistant	47	NR	55	Fatigue, myalgia	[28]
Brigatinib	Phase 1/2	Both	24	NR	62.5	Fatigue, nausea, diarrhea	[33]
PF-06463922	Phase I/2	Both	15*	NR	40%	Hypercholesteremia, peripheral edema	[37]

*Included: 18 ALK and 4 ROS patients, 19 pretreated with crizotinib/ceritinib and 17 patients having a CNS metastasis.

When comparing alternative ALK inhibitors to ceritinib, several things stand out. Firstly, alectinib has a potency to inhibit ALK in the same range as ceritinib (52 nM vs. 25 nM respectively). Secondly, alectinib is able to inhibit two of the most common crizotinib-resistant mutations in preclinical models (L1196M and G1269A), similar to ceritinib.[38] However, alectinib can also inhibit three distinctly different ALK mutants that ceritinib cannot (C1156Y, F1174L, R1275Q) (Table 3). As noted before, F1147L is also an ALK mutation that confers resistance to crizotinib, indicating a benefit for alectinib but not for ceritinib. Both ceritinib and alectinib have similar ORRs in crizotinib-resistant NSCLC patients (58% vs. 59% respectively) (Table 4). However, significant differences can be seen in the disease control rate (DCR) for these two ALK inhibitors in relation to CNS lesions after crizotinib treatment in NSCLC patients. The DCR of CNS lesions in patients receiving crizotinib was 56% at 24 weeks. For alectinib, the DCR for CNS lesions in crizotinib-treated NSCLC patients ranged from 77.8% to 100% while ceritinib showed a similar DCR of 84.8% for CNS lesions after crizotinib treatment. However, the intracranial response rates for alectinib and ceritinib differ (alectinib 55.6%–68.8%; ceritinib 39.4%). These results imply that alectinib is more effective at inhibiting CNS lesions after crizotinib resistance occurs.[39] The differences between these drugs can be explained by the poor penetration of crizotinib into the brain, because of an efficient P-gP-mediated efflux, while both ceritinib and alectinib seem to be poor substrates for P-gP.

The clinical trial for alectinib is expected to be finalized by the end of September 2015, which will elucidate the progression-free survival of this group of patients, hopefully revealing the efficacy of alectinib as treatment for crizotinib-resistant NSCLC tumors.[40] Until now, both ceritinib and alectinib seem similarly effective treatments, albeit toward different target mutations. Additional biopsy of recurrent tumors in order to identify the exact mechanism of acquired resistance is needed to determine the most effective therapy.

Less is known about the efficacy of brigatinib in crizotinib-resistant NSCLC patients. Brigatinib has a similar selectivity for ALK as ceritinib in cellular assays (21 nM vs. 25 nM) but is less effective against crizotinib-resistant ALK-mutants. The only known ALK target mutation that brigatinib is able to inhibit is the gatekeeper mutation L1196M, which is also inhibited by ceritinib. The difference between the two ALK inhibitors is that brigatinib also inhibits EGFR gatekeeper mutation T790M. Up to 60% of NSCLC patients that were treated with EGFR tyrosine kinase inhibitors acquired a T790M mutation, conferring resistance to EGFR inhibition.[44,45] No overlap can be seen among 103 ALK-positive patients and 214 EGFR-mutation positive patients, implying that EGFR activation in crizotinib-naïve ALK-positive NSCLC is not due to EGFR mutation.[15] Together, these results imply that brigatinib can be useful in patients that not only are resistant to crizotinib due to the ALK gatekeeper mutation L1196M, but also in patients with acquired resistance to EGFR inhibition by T790M. Resistance to ALK inhibition by crizotinib has been shown to occur by EGFR activation in vitro and additional EGFR inhibition was shown to be beneficial in limiting tumor growth.[46] If in these cases, secondary acquired resistance occurs through either ALK L1196M gatekeeper mutations, or EGFR T790M mutations, and then brigatinib could be an effective therapy.

Expert commentary

The efficacy of ceritinib in crizotinib-resistant ALK-positive NSCLC patients is based on well-researched mechanisms and proven to be a solid choice as therapy for recurrent NSCLC. The safety profile of ceritinib does suggest increased toxicity compared to crizotinib but this effect could be explained by the different dose regimen of the drugs (750 mg once daily ceritinib, 250 mg twice daily crizotinib). Over 60% of the patients received a dose reduction due to adverse effects of ceritinib treatment. A second dose-optimizing study is currently being performed by Novartis in order to minimize toxicity while maintaining optimal effectiveness. A possible food effect on the exposure to ceritinib needs further investigation as well.

Although distinct differences have been made to the molecular structure of crizotinib in order to design ceritinib, their pharmacokinetic profiles look similar. It is therefore not unlikely that the above-mentioned dose-optimizing study will result in an adjusted (lower) recommended dose. Interestingly, crizotinib was originally designed as a c-MET inhibitor. C-MET has similar downstream targets as ALK such as PI3K and RAS/RAF/ERK allowing crizotinib to effectively inhibit two pathways that contribute to oncogenesis of NSCLC tumors by increasing proliferation and evade cell death. C-MET amplification is present in approximately 7% of NSCLC patients [47] but c-MET amplification and ALK-translocation in NSCLC are mutually exclusive.[48] In addition, ALK-rearranged NSCLC never seems to harbor other oncogenic driver mutations such as EGFR or RAS mutations. Therefore, ALK-rearranged NSCLC is often ALK-dependent, supporting the efficacy of crizotinib.[49] These results suggest that although crizotinib is highly effective at inhibiting these ALK-driven tumors, ceritinib might be even more effective. A phase 3 trial is currently undergoing to assess the efficacy of ceritinib as a first-line treatment in ALK-rearranged NSCLC (Identifier: NCT01828099).

Currently, ceritinib is the only FDA-approved second-generation ALK-inhibitor. Ceritinib has a low IC50 for growth inhibition (25 nM) in Ba/F3-ALK models, targets the two most common crizotinib-resistant mutants LL119M and G1269A, and has been tested in the largest patient population (172 ALK-positive NSCLC patients). What is interesting about the competing next-generation ALK-inhibitors are the differences in target mutations. While ceritinib targets the two most common crizotinib-resistant ALK mutants, there are, and will be, cases where ceritinib treatment will not be the best choice. In addition, a recent study further supported the significance of ALK-mutant identification. This case report showed that the ALK-mutant F1174V confers sensitivity to alectinib while the ALK-mutant I1171 confers resistance.[50]To conclude, genetic identification of the crizotinib-resistant mutants will allow doctors to effectively choose the optimal treatment strategy for a single patient, thus solidifying the importance of actively identifying driver mutations or aberrations in cancer patients by genotyping and personalized medicine.

Five-year view

In addition to ceritinib, other second-generation ALK inhibitors are currently in development and undergoing FDA approval. The existence of multiple effective ALK inhibitors, each with their own specific ALK-mutation targets, offers a much sought-after approach to cancer medicine. Identification of the exact mutations that confer resistance to crizotinib and the drugs that can overcome them gives clinicians the tools to accurately battle recurrent tumors in ALK-rearranged NSCLC. The importance of this sequential method of treatment is further supported by a recent study that reported a superior effect of second line ALK-TKI (either ceritinib or alectinib) after first-line ALK-TKI (mostly crizotinib) compared to patients who transitioned to other systemic treatments (P = 0.03) [51] As mentioned before, the ALK-inhibitor named PF-06463922 is currently being developed by Pfizer©. This ALK/ROS1 inhibitor can effectively inhibit all clinically available crizotinib-resistant ALK-mutants in vitro (including the highly resistant G1202R mutation).[31] In addition, it was designed specifically to overcome the low penetrance of crizotinib in the CNS. In the ASCEND 2 (chemotherapy and crizotinib pre-treated) and 3 (chemotherapy and ALK inhibitor naïve) studies, ORR were 38.6% and 63.7%, respectively, while for patients with brain metastases these values were almost comparable (33% and 58 %, respectively).[42,43] All together, these results show that a highly CNS-penetrant and effective treatment is feasible for all ALK-rearranged NSCLC patients that become resistant to crizotinib.

The principle of personalized medicine has been proven effective, and knowledge of other escape mechanisms, such as increased EGFR-signaling (occurring in approximately 30% of crizotinib-resistant NSCLC patients) and KIT amplification, gives even more targets for the development of scientifically validated inhibitors.[52]

Combination treatments of EGFR tyrosine kinase inhibitors and ALK inhibitors are currently being investigated in order to characterize the effect of dual treatment strategies. The situation where a patient will receive more than two or three cancer drugs in succession due to the identification and targeting of escape mechanisms is becoming more and more plausible. Recently, a study by Nie et al. demonstrated how cell-free circulating DNA could be used to genotype NSCLC tumors.[53] These results suggest that non-invasive testing of blood samples from NSCLC patients could be used to accurately genotype the recurrent tumors, thereby identifying the mutations responsible for recurrence. Also, monitoring these patients throughout their treatment by repeatedly testing for crizotinib-resistant ALK DNA in their blood plasma could help to anticipate recurrence. Hopefully, in the near future, these strategies would be more cost-effective than they are today, especially, considering the small population of ALK-rearranged NSCLC patients, and the even smaller population of patients harboring crizotinib-resistant ALK mutants.

Key issues

• The poor pharmacokinetic profiles of some anaplastic lymphoma kinase (ALK) inhibitors may limit their functionality as ALK inhibitor. The low solubility and the limited capacity to penetrate the CNS impair the overall efficacy of these drugs significantly. Inhibition of the p-glycoprotein efflux pump activity could also be used to increase CNS penetrance of these molecules and prevent CNS recurrence.

• Intracellular sequestration in the lysosomes of various TKI including the ALK inhibitor crizotinib limits exposure of the target and subsequently efficacy of these drugs.[24]

• Overcoming these hurdles in the next-generation ALK inhibitors seems of vital importance.

• Combination therapies to improve the efficacy of these drugs will provide valuable alternatives. For example, enhancing the effect of ALK inhibition by combination treatment with an Hsp90 inhibitor, impairing the folding and stability of EML4-ALK by Hsp90, might be a viable strategy.[54]

• However, these strategies are only viable if the added value of widespread genotyping of all non-small cell lung cancer patients exceeds the costs.

Financial & competing interests disclosure

The author has no 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. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.