Personalized Medicine Tackles Clinical Resistance: Alectinib in ALK-Positive Non–Small Cell Lung Cancer Progressing on First-Generation ALK Inhibitor

In 2007, the identification of rearrangements involving ALK, a gene encoding a transmembrane receptor tyrosine kinase, in 3% to 7% of non–small cell lung cancer (NSCLC) series enriched in adenocarcinoma histology heralded a second major breakthrough in the field of precision therapy for molecularly defined subsets of NSCLC (1, 2). Among several identified 5′ fusion partners, the echinoderm microtubule–associated protein-like 4 (EML4)-ALK fusion variant 1, a product of an inversion genetic event in the short arm of chromosome 2, represents the most common ALK fusion in patients with NSCLC. It promotes constitutive expression and ligand-independent oligomerization of the fusion protein, thus resulting in its aberrant activation. ALK translocation is observed predominantly in younger patients with adenocarcinoma histology—typically never or former light smokers—and has been associated with high frequency of metastatic spread to the liver, pericardium, and pleura (3). Monotherapy with crizotinib (Xalkori; Pfizer), an orally bioavailable ATP-competitive inhibitor of the ALK, ROS1 (ROS proto-oncogene 1, receptor tyrosine kinase), and MET (MET proto-oncogene, receptor tyrosine kinase) tyrosine kinases, at a dose of 250 mg orally twice daily, represents the current standard of care in the first-line treatment of ALK-positive, locally advanced or metastatic NSCLC on the basis of demonstrated superior efficacy compared with a platinum/pemetrexed doublet in the PROFILE 1014 phase III clinical trial. In the intention-to-treat population, crizotinib resulted in an objective response rate [ORR; by independent review committee (IRC)] in the crizotinib arm of 74% [95% confidence interval (CI), 67–81] compared with 45% (95% CI, 37–53) in the chemotherapy arm (P < 0.001) with median progression-free survival (PFS) of 10.9 months (95% CI, 8.3–13.9) and 7.0 months (95% CI, 6.8–8.2), respectively (HR for progression or death with crizotinib, 0.45; 95% CI, 0.35–0.60; P < 0.001). The superiority of crizotinib compared with chemotherapy was further documented in the context of platinum-refractory disease in the PROFILE 1007 study. Comparable antitumor activity was previously reported in the PROFILE 1001 phase I and PROFILE 1005 single-arm phase II studies that supported the drug's accelerated approval by the FDA in 2011 (4).

Despite favorable initial response to crizotinib in the majority of ALK-rearranged patients with NLSLC, relapse is inevitable and, in most cases, occurs within the first year after initiation of therapy. The central nervous system (CNS) represents a hotspot for relapse/progression and is frequently the single site of crizotinib failure despite documented activity of the drug in brain metastases that can be further enhanced with radiotherapy (5–8). This tropism is likely attributable to poor CNS drug penetration that results in establishment of a “sanctuary site” for expansion of metastatic clones. Both limited passive diffusion across the blood–brain barrier and active, P-glycoprotein–mediated efflux account for the reduced intracerebral concentration of crizotinib, with a cerebrospinal fluid (CSF)/plasma ratio of 0.0026 recorded from a patient treated at the recommended dose of 250 mg orally twice daily (7, 9). In cases of extracranial, systemic progression, secondary mutations in the ALK tyrosine kinase domain that reduce the binding affinity of crizotinib for the mutant oncoprotein have been identified in 20% to 36% of patients, with copy number gain of the ALK fusion gene observed in 8% to 10% of patients (10–13). Contrary to the dominance of T790M among secondary EGFR mutations conferring acquired resistance to first-generation EGFR tyrosine kinase inhibitors (TKI), a number of different resistance mutations in the ALK tyrosine kinase domain have been identified. These include the prevalent L1196M gatekeeper (identified in ∼7% of crizotinib-resistant tumors) and G1269A mutations (identified in ∼4% of crizotinib-resistant tumors), as well as the lower frequency L1152R, C1156Y, 1151 T-ins, I1171T, and F1174L mutations and the solvent front G1202R and S1206Y mutations (2, 10–15). Nonetheless, in the majority of cases of clinical resistance to crizotinib, a secondary genetic alteration in ALK cannot be identified; rather, activation of bypass tracks mediates complete or partial loss of dependence on ALK kinase activity for activation of downstream signal transduction pathways. Identified mechanisms include amplification of KIT (KIT proto-oncogene receptor tyrosine kinase) coupled with increased expression of its ligand–stem cell factor (SCF)–from stromal cells; activation of EGFR, SRC (SRC proto-oncogene, non-receptor tyrosine kinase), protein kinase C, or IGF1R (insulin-like growth factor 1 receptor)-mediated signaling; as well as de novo acquisition of activating mutations in EGFR and KRAS (KRAS proto-oncogene, GTPase) or expansion of preexisting mutant clones (11, 12, 16–18). Other mechanisms of MAPK pathway activation, including wild-type KRAS copy number gain and reduced expression of DUSP6 (dual specificity phosphatase 6), have also been reported (19). In several cases, more than one resistance mechanism can coexist in the same patient (11). Finally, both transformation to small cell lung cancer and acquisition of a mesenchymal phenotype have been associated with acquired resistance to crizotinib, in accordance with similar reports in EGFR-mutant NSCLC at the time of acquired resistance to EGFR TKIs (20, 21).

To tackle acquired biologic resistance to crizotinib as well as pharmacokinetic failure in the CNS, a number of highly potent, second-generation ALK inhibitors were developed. In December 2015, alectinib (Alecensa; Roche/Genentech) became the second inhibitor in this category to gain accelerated approval by the FDA for the treatment of patients with ALK-positive (as determined by an FDA-approved test), metastatic NSCLC who have progressed on or are intolerant to crizotinib, following the earlier approval of ceritinib (Zykadia; Novartis) for the same indication in April 2014.

Alectinib was developed by Chugai Pharmaceutical as a potent and highly selective, ATP-competitive, orally bioavailable ALK TKI. In cell-free enzymatic assays, its IC₅₀ for ALK kinase activity was 1.9 nmol/L (22). The compound further displayed remarkable selectivity, with recorded IC₅₀ values ≤10 nmol/L for only two other kinases, leukocyte receptor tyrosine kinase (LTK) and cyclin G–associated kinase (GAK; ref. 22). This activity profile translated into robust antiproliferative and proapoptotic effects against NSCLC, anaplastic large-cell lymphoma, and neuroblastoma cell lines harboring ALK aberrations. There was further evidence of in vivo efficacy in the H2228 NSCLC xenograft model where once-daily oral administration of alectinib at a dose of either 20 or 60 mg/kg induced substantial and sustained tumor regression without overt toxicity (22). Importantly, alectinib exhibited strong binding affinity for ALK bearing the gatekeeper L1196M resistance mutation (K_i = 0.83 nmol/L for wild-type ALK; K_i = 1.56 nmol/L for L1196M) and was active in both cellular and xenograft models of L1196M (22). Comparable inhibitory activity to wild-type ALK was also demonstrated against mutant proteins bearing the C1156Y and F1174L resistance mutations, and translated to potent antiproliferative activity in cellular models (22).

The accelerated FDA approval of alectinib was based on efficacy data from two single-arm phase II clinical trials: NP28761 (NCT01871805), a study conducted across centers in the United States and Canada, and NP28673 (NCT01801111), a global study conducted across 16 countries in Europe, Asia, North America, and Australia (23, 24). Both studies enrolled patients with locally advanced or metastatic, ALK-positive NSCLC (as determined by an FDA-approved FISH test) with Eastern Cooperative Oncology Group (ECOG) performance status (PS) 0 to 2 who had progressed on crizotinib. Importantly, both trials included patients with treated and stable or untreated asymptomatic brain metastases and were, therefore, representative of a “real world” cohort of patients refractory to crizotinib. Indeed, 60% and 61% of patients, respectively, had evidence of CNS metastases at trial entry. Prior chemotherapy was allowed in both studies.

Updated results from both studies were recently published and confirmed robust activity of alectinib in the crizotinib-refractory setting. In NP2876 (N = 87), the confirmed ORR by IRC across 69 response-evaluable patients was 52% (95% CI, 40–65), with a median duration of response (DoR) of 13.5 months (95% CI, 6.7–NE) and median PFS of 8.1 months (95% CI, 6.2–12.6) in the intention-to-treat population (25). Similar results were reported in NP28673, with a confirmed ORR of 50% (95% CI, 41–59) among 122 response-evaluable patients, median DoR of 11.2 months (95% CI, 9.6–NR), and median PFS of 8.9 months (95% CI, 5.6–11.3) for all 138 patients enrolled in the study (23).

Critically, in both studies, alectinib demonstrated significant activity in the CNS. Among 16 patients with measurable CNS disease in NP28761, an objective response (by IRC) was obtained in 75%, with 25% of all patients achieving a complete CNS response (24). In NP28673, the ORR among 35 patients with measurable CNS disease was 57% (95% CI, 39–74), with complete CNS response observed in 27% of the entire cohort of patients with measurable and nonmeasurable CNS disease (23). Across both trials, CNS responses occurred regardless of prior CNS radiotherapy and were durable with median CNS DoR of 11.1 months (95% CI, 5.8–11.1) and 10.3 months (95% CI, 7.6–11.2), respectively. Although no patients in these studies were reported to have leptomeningeal disease, separate reports have confirmed that alectinib is active in this setting (26).

Robust systemic and CNS efficacy of alectinib was further demonstrated in the first-line setting in the single-arm phase I/II AF-001JP (JapicCTI-101264), conducted in Japan. ALK translocation status was assessed by immunohistochemistry in this study and subsequently confirmed by either FISH or RT-PCR. In the phase II component of the trial, among 46 patients treated at the recommended phase II dose of 300 mg orally twice daily, the ORR by IRC was 93.5% (95% CI, 82.1–98.6), with a median PFS of 28 months (27).

The toxicity profile of alectinib is favorable compared with crizotinib and ceritinib, with a markedly reduced rate of gastrointestinal toxicities and a low rate (6%) of permanent adverse event–related treatment discontinuation. Among 253 patients treated with 600 mg per os twice daily, the most common adverse reactions were fatigue (41% all grades; 1.2% grades 3/4), constipation (34% all grades; 0% grades 3/4), edema (30% all grades; 0.8% grades 3/4), myalgia (29% all grades; 1.2% grades 3/4; ref. 28). Common laboratory abnormalities included increased aspartate aminotransferase (AST; 51% all grades; 3.6% grades 3/4), increased alanine aminotransferase (ALT; 34% all grades; 4.8% grades 3/4), increased alkaline phosphatase (47% all grades; 1.2% grades 3/4), increased creatine phosphokinase (CPK; 43% all grades; 4.6% grades 3/4), and hyperbilirubinemia (39% all grades; 2.4% grades 3/4; ref. 28). In the majority of cases, these were reversible following temporary discontinuation of treatment with or without dose reduction. The most frequent adverse events leading to treatment discontinuation were hyperbilirubinemia and increased ALT/AST. Monitoring of liver function tests every 2 weeks during the first 2 months of therapy and of CPK every 2 weeks for the first month of therapy is recommended. Other label warnings include interstitial lung disease/pneumonitis (0.4% grade 3), bradycardia, and fetal toxicity. Photosensitivity has been reported in 9.9% of patients, and patients are advised to avoid sun exposure and to use sunscreen while on therapy (28).

The absorption of alectinib increases when it is administered within 30 minutes of a meal. Under fed conditions, the median time to C_max is 4 hours, and the mean single dose half-life is approximately 20 hours (28). Alectinib is metabolized by the CYP3A4 (cytochrome P450 family 3 subfamily A member 4) hepatic microsomal system primarily to M4, an active metabolite (also a CYP3A4 substrate) with comparable in vitro potency to the intact drug (28). Accordingly, coadministration of CYP3A4 inducers or inhibitors does not substantially affect the combined exposure to alectinib and M4. It is primarily excreted in the feces. No dose adjustments are recommended in the context of mild-to-moderate renal impairment (creatinine clearance ≥30 mL/minute) or mild hepatic impairment. In keeping with its potent CNS activity, concentrations in the CSF in patients with ALK-positive NSCLC are comparable to those of unbound drug in the plasma and the drug is not a substrate for the ABCB1 (ATP-binding cassette subfamily B member 1; P-glycoprotein) efflux pump (29).

The full spectrum of different mechanisms of clinical resistance to alectinib has been incompletely characterized, yet it is evident that both ALK-dependent and ALK-independent (signaling bypass) mechanisms are in operation, analogous to what has been described previously for crizotinib and ceritinib. In keeping with the increased potency of alectinib compared with crizotinib, the incidence of secondary ALK resistance mutations is significantly higher for alectinib (53%) than for crizotinib (20%) and mirrors that observed following treatment with ceritinib (54%; ref. 13). It is also noteworthy that due to differences in the structure of these inhibitors, emerging resistance mutations result in partially nonoverlapping patterns of sensitivity to second generation ALK TKIs, thereby posing opportunities for sequential therapy (Table 1; refs. 10–13, 15, 22, 25, 30–46). Mutations involving I1171 in the hydrophobic regulatory spine of the ALK kinase domain (I1171T/N/S) have been identified in 12% to 56% of patients with available sequencing data following failure of alectinib in two separate series (13, 41). Molecular simulation analysis indicated that the I1171T mutation is predicted to disrupt a hydrogen bond between E1167 and alectinib, thus impairing drug binding. In vitro in the H3122 ALK-rearranged NSCLC cell line, acquired high-level resistance to alectinib could be induced by a secondary V1180L mutation in the ALK kinase domain, and this was recently confirmed in a patient who developed resistance to alectinib (13). Computational modeling indicated steric clash between the methyl group of the leucine residue and the multicyclic rings of alectinib as the likely mechanism, resulting in reduced binding affinity of the drug for the mutated kinase (47). Importantly, both V1180L and I1171T mutants are sensitive to ceritinib in vitro. In the case of I1171T, this has been further confirmed in vivo in an alectinib-resistant patient who responded to subsequent treatment with ceritinib (47). In contrast, the solvent front G1202R mutation, identified in 22% to 29% of alectinib-resistant tumors with available sequencing data, confers high-level resistance to alectinib, ceritinib, and crizotinib, but not to lorlatinib (PF-06463922; refs. 13, 41, 48).

Activation of signaling bypass tracks that promote ALK independence has also been observed in alectinib-resistant tumors and cellular models, although the true incidence of this resistance mechanism in vivo is currently unknown. MET amplification was noted in an alectinib-resistant patient who subsequently responded to crizotinib (49), and MET activation via an HGF-mediated autocrine feedback loop was further identified in a cellular model of acquired resistance to alectinib that retained sensitivity to crizotinib (50). In vitro, acquired resistance to alectinib in the ALK translocation–positive H3122 NSCLC cell line was mediated by EGFR pathway activation via an NRG1 (neuregulin 1)–ERBB33 (erb-b2 receptor tyrosine kinase 3)–EGFR axis or a TGFα–EGFR autocrine loop, and could be reversed by cotreatment with afatinib, an irreversible second-generation EGFR TKI (51, 52). Coactivation of the NRG1–ERBB3 and IGF1R pathways, with concomitant loss of the EML4–ALK oncogene in the H2228 ALK-rearranged NSCLC cell line, conferred ALK independence and alectinib resistance, but suppression of downstream signaling and reduction of cell viability could be achieved with combined inhibition of EGFR and IGF1R (50).

Finally, in stark similarity to patterns of resistance to EGFR TKIs, morphologic transformation to small cell carcinoma and epithelial–mesenchymal transition have also been reported in patients who developed resistance to alectinib (13, 53).

In addition to alectinib, ceritinib has received FDA approval for the treatment of patients with ALK-positive NSCLC whose disease has progressed during treatment with crizotinib or who have became intolerant to the drug. Six other next-generation ALK TKIs are undergoing early-phase clinical development in this space including brigatinib (AP26113; ARIAD), lorlatinib (PF-06463922; Pfizer), entrectinib (RXDX-101, NMS-E628; Ignyta), TSR-011 (TESARO), X-376/X-396 (Xcovery), and CEP-28122/CEP-37440 (Teva; Table 2; ref. 54). Clinical development of a ninth inhibitor, ASP3026 (Astellas Pharma), was suspended for strategic reasons.

In the global phase I study of ceritinib (ASCEND-1; NCT01283516), the ORR among 163 ALK TKI-pretreated patients was 56% (95% CI, 49–64), with a median DoR of 8.3 months (95% CI, 6.8–9.7) and median PFS of 6.9 months (95% CI, 5.6–8.7). Disease control in the CNS was reported in 65% (95% CI, 54–76) of patients (45). In the first-line setting, the ORR to ceritinib was 72% (95% CI, 61–82), with a median DoR of 17 months (95% CI, 11.3–NE) and median PFS of 18.4 months (95% CI, 11.1–NE; ref. 55).

Preliminary efficacy and safety data from the phase I clinical trial of lorlatinib (PF-06463922; NCT01970865) were reported during the 2015 World Conference on Lung Cancer. In a patient cohort that included both treatment-naïve and ALK TKI–refractory patients, several of which had progressed on more than one ALK TKI, the ORR (including both confirmed and unconfirmed responses) was 47% (95% CI, 31–62; ref. 56). Importantly, partial response to lorlatinib was observed in a patient bearing the G1202R mutation, thus confirming preclinical data of lorlatinib activity against this recalcitrant mutation (48).

In the single-arm phase I/II trial of brigatinib (AP26113; NCT01449461), among 70 patients with ALK-positive advanced NSCLC whose disease had progressed on prior crizotinib, the ORR was 71% (95% CI, 59–82) at the February 17, 2015, data cutoff point, with a median PFS of 13.4 months (57). Substantial intracranial activity was observed, with 53% ORR recorded among patients with measurable intracranial metastases and 35% among patients with nonmeasurable intracranial metastases. Of note, pulmonary toxicity occurred in 14% of patients who were treated with 180 mg once daily and usually occurred during the first week of treatment. A more gradual increase in the dose to 180 mg once daily after 1 week at the 90-mg once-daily dose level substantially improved tolerance. Further efficacy results from the ongoing, randomized phase II ALTA clinical trial (NCT02094573) are awaited.

In the absence of data from randomized head-to-head trials comparing second-generation ALK TKIs and comparable reported efficacy measures of alectinib and ceritinib in their respective single-arm studies, the choice of second-line ALK TKI following crizotinib failure may require a personalized treatment approach. The robust activity of alectinib against CNS disease and its favorable toxicity profile may favor its use in this context. Differential activity toward specific resistance mutations may also influence choice of ALK TKI. Although both drugs are active against the most common crizotinib resistance mutations, L1196M and G1269A, alectinib retains activity against the F1174L/V ceritinib resistance mutations. Conversely, ceritinib is active against the prevalent I1171T/N/S alectinib resistance mutation. Thus, sequential use of different second-generation ALK TKIs, guided by molecular profiling of resistant tumors, may represent a viable therapeutic strategy for the subset of patients whose disease remains ALK dependent. As noted previously, the G1202R mutation confers resistance to both alectinib and ceritinib, but may respond to treatment with the third-generation inhibitor lorlatinib (PF-06463922).

Mirroring current controversies in the treatment of EGFR-mutant NSCLC, the optimal sequencing of ALK TKIs for patients with ALK rearrangement–positive advanced NSCLC has not yet been determined. The global ALEX randomized phase III clinical trial (NCT02075840) and similar J-ALEX randomized phase III trial in Japan are comparing crizotinib with alectinib for the first-line therapy of patients with ALK-positive NSCLC. In February 2016, it was announced that J-ALEX was stopped prematurely because the study met its primary endpoint at a preplanned interim analysis. Because of the lack of crossover at the time of progression in the design of these trials, direct comparison of upfront treatment with alectinib versus crizotinib as first-line therapy followed by alectinib at the time of progression will not be possible. More direct evaluation of the optimal sequencing of first- and second-generation ALK TKIs will be provided by the planned NCI ALK Master Protocol that will compare crizotinib with different second-generation ALK inhibitors as first-line therapy, with crossover allowed for each arm of the study at the time of progression.

The development of rational combinatorial strategies with ALK TKIs is also underway. The combination of alectinib with atezolizumab (MPDL3280A), a PD-L1 inhibitor, is being assessed in a phase Ib clinical trial (NCT02013219) in patients with treatment-naïve, ALK-positive NSCLC. In addition, the combination of alectinib with bevacizumab is being assessed in a phase I/II clinical trial (NCT02521051).

V.A. Papadimitrakopoulou reports receiving commercial research grants from AstraZeneca, Bayer, Bristol-Myers Squibb, Celgene, Clovis Oncology, Genentech, Janssen, Merck, Novartis, and Pfizer, and is a consultant/advisory board member for ARIAD, AstraZeneca, Genentech, Janssen, and Merck. No potential conflicts of interest were disclosed by the other author.

Conception and design: F. Skoulidis, V.A. Papadimitrakopoulou

Development of methodology: F. Skoulidis

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Skoulidis, V.A. Papadimitrakopoulou

Writing, review, and/or revision of the manuscript: F. Skoulidis, V.A. Papadimitrakopoulou

F. Skoulidis is supported by the Andrew Sabin Family Foundation, the Lung Cancer Research Foundation, and a Sheikh Khalifa Bin Zayed Al Nahyan Scholar Award.