The major recommendations, by the International Association for the Study of Lung Cancer (IASLC), are to use testing for ALK fusions to guide patient selection for therapy with an ALK inhibitor in all patients with advanced-stage adenocarcinoma, regardless of sex, race, smoking history or other clinical risk factors (21). Of importance is the method which should be used for ALK testing. To date, the fluorescent in situ hybridization (FISH) assay using dual-labelled break-apart probes for selecting patients for ALK-TKI therapy is the diagnostic gold standard approved also by FDA (21). This methodology was used in the initial studies that demonstrated improved clinical response of patients with ALK-rearranged tumours to treatment with crizotinib, an ALK-TKI (12,13,22). As reported by this test, the indicative cut-off of an ALK rearrangement is the observation of >15% split nuclei (22). ALK immunohistochemistry (IHC) being an easier and cheaper diagnostic tool is currently under investigation to be considered as a pre-screening methodology to select specimens for ALK FISH testing (21). Until now, many reports demonstrated a strong correlation between ALK IHC expression and ALK FISH positivity or negativity (23–27); however, further confirmation is needed to definitely include the ALK IHC detection as the initial step in the algorithm for clinical ALK testing in NSCLC. Reverse transcriptase polymerase chain reaction (RT-PCR) is not currently recommended as a first-line diagnostic method for determining ALK fusion status. A higher failure rate of a ribonucleic acid (RNA)-based assay in routine formalin fixed paraffin-embedded (FFPE) pathology material might affect sensitivity, with the risk of false-negatives, due to variability in the EML4-ALK fusion structure and to the existence of other ALK fusion partners (21).

The first evidence of mechanisms of resistance to crizotinib were described for a 28-year-old man, never smoker with advanced pretreated EGFR wild-type and ALK-positive lung adenocarcinoma who relapsed after crizotinib therapy and was biopsied at progression. In this patient, two secondary point mutations (L1196M and C1156Y) within the kinase domain of the EML4-ALK were described, each mutation developing independently in subclones of the tumour and conferring marked resistance to two different ALK inhibitors (39). Furthermore, several reports showed that in approximately one-third of relapsing patients, the crizotinib resistance is mediated by the appearance of mutations located in the ALK TK domain. The most commonly identified resistance mutation is the ‘gatekeeper mutation’ L1196M which hinders crizotinib binding at its active site on ALK kinase (40–45). Some series reported that the amplification of the ALK fusion gene with or without concurrent ALK mutation leads to drug resistance (43,44).

One of the potential new pathway activation, which might mediate crizotinib resistance is that of EGFR signalling. In this setting, a double inhibition, of ALK and EGFR, was effective (40,43). Activation of KRAS might represent another resistance mechanism in ALK-positive tumours due to the detection of new KRAS mutations in ALK-positive NSCLC resistant to crizotinib (43).

An analysis performed on 16 ALK-positive NSCLC patients treated with crizotinib and re-biopsied at progression, suggested that crizotinib-resistant patients might be subdivided into two main groups. The ALK dominant group, which represents approximately 50% of cases and includes ALK kinase mutation (31%) and ALK fusion gene copy number gain (13%), and the ALK non-dominant group, which represents the other half, including emergence of alternate EGFR or KRAS mutations (31%) and in 19% of cases in which the oncogene remained unknown and could rely on KIT, EGFR or HER-2 variants (46).

If the crizotinib resistance is mediated by ALK mutation or amplification it means that the NSCLC is still addicted to ALK signalling: this is the case in which a second generation ALK inhibitors might represent a promising treatment approach. These new ALK inhibitors are structurally distinct from crizotinib and with the capability to inhibit also secondary acquired mutations (40,43). Among the most promising next generation ALK inhibitors, LDK378 is active in conventional ALK-positive tumours and in those expressing the mutation C1156Y. In a dose-escalating phase I single-arm study, the maximum tolerated dose, safety, pharmacokinetics and preliminary antitumour activity of LDK378 were investigated in ALK-positive solid tumours of any type. A total of 114 patients had ALK-positive NSCLC, 78 patients taking LDK378 at 750 mg/day and 36 additional patients taking 400–750 mg/day. All 114 patients were evaluable for response, with 79 patients progressing during or following treatment with crizotinib, and 35 patients were crizotinib-naïve. The objective response rate was 60% in the 78 patients who received LDK378 at 750 mg/day. In the 114 patients treated with LDK378 at doses ≥400 mg/day, the objective response rate was 58%, with a median duration of response of 8.2 months and a median PFS of 8.6 months. The most frequent adverse events were nausea (73%), diarrhoea (72%), vomiting (58%) and fatigue (41%). The most frequent grade 3–4 adverse events were increased alanine aminotransferase (19%), increased aspartate aminotransferase (10%) and diarrhoea (8%) (48). These very impressive results led to the design of further trials which are actively recruiting worldwide. Among these a phase II multicenter, open label, single-arm study is evaluating the efficacy and safety of oral LDK378 750 mg once-daily in patients with ALK-positive advanced NSCLC who received 1–3 lines of therapy (including 1 platinum doublet) and progressed on crizotinib as the last therapy prior to study entry. The study was designed to enrol 137 patients (49).

Alectinib (CH5424802) was a potent, selective, ALK inhibitor with 10-fold more potency than crizotinib and effectives against most of the mutations of the ALK domain. A multicentre, single-arm, open-label, phase I/II study, ALK-rearranged inhibitor-naïve advanced NSCLC patients received alectinib orally twice daily by dose escalation. In the phase I study, 24 patients were treated with 20–300 mg twice daily. No dose limiting toxicities or grade 4 adverse events were reported, even with the highest dose. Thus, 300 mg twice daily was the recommended dose for the phase II study. A total of 46 patients were treated with the recommended dose, with an objective response rate of 93.5% including two complete responses (4.3%). Grade 3 treatment-related adverse events were reported in 12 (26%) patients, serious side effects occurred in 5 patients (11%). Median PFS has not been reached yet, since 40 of the 46 patients in the phase II portion remain on treatment (50). Another phase I study enrolled 47 ALK-positive NSCLC patients who failed crizotinib and chemotherapy. No dose limiting toxicities were observed up to the highest dose tested (900 mg twice daily). Grade 3–4 adverse events included γ-glutamyl transferase increase (n=2), neutrophil decrease (n=2), hypophosphatemia (n=2), hyperglycemia, syncope, renal failure and pericardial effusion (n=1 each), but no grade 3 nausea, vomit, diarrhoea, oedema were reported. The objective response rate was 54.5% across all dose cohorts for all patients, all partial responses, with a median duration on treatment >4 months. Alectinib at the oral dose of 600 mg twice daily was chosen as recommended phase II dose based on pharmacokinetics, efficacy and tolerability (51). In both trials, the promising antitumour activity of alectinib was observed in ALK-positive patients including also those who failed crizotinib. Therefore, a global single-arm phase II study of alectinib in crizotinib-resistant ALK-positive NSCLC patients is ongoing (52) (Table III).

AP26113 is a novel dual ALK-/EGFR-TKI that potently inhibits mutated forms including the ALK L1196M and EGFR T790M mutations. A phase I/II open-label, multicenter study enrolled 44 patients with advanced solid tumours, including 37 patients with NSCLC, refractory to available therapies or for whom no standard treatment exists. AP26113 was administered with escalating once-daily oral doses. Most common 3–4 treatment-related adverse event was diarrhoea, reported in 5% of cases. Two dose limiting toxicities were observed at 240 mg (grade 3 alanine aminotransferase increase) and at 300 mg (grade 4 dyspnea). Thus, doses <300 mg are being explored further in the ongoing phase II expansion, which includes 4 cohorts: ALK-positive NSCLC naïve or resistant to prior ALK-targeted therapy, EGFR mutated NSCLC resistant to EGFR-targeted therapy, other cancers with abnormalities in ALK or other AP26113 targets (53).

ALK fusion proteins, including those with resistance mutations, are known heat shock protein 90 (HSP90) clients. HSP90 inhibitors bind in the ATP-binding pocket of the enzyme, and prevent it from regulating the activation and stability of its client proteins, including ALK. Thus, inhibition of HSP90 resulted in reduction of the expression of EML4-ALK through proteasome-mediated degradation (41,54). Unfortunately, to date no clinical trial has been conducted to investigate HSP90 inhibitors specifically addressed to patients with ALK-positive NSCLC, but retrospective analysis showed promising results of HSP90 inhibitors in this setting (55–57). Ganetespib (STA-9090) an HSP90 inhibitor, was administered as monotherapy at the dose of 200 mg/m² weekly, for 3 weeks with a one-week rest, to 99 patients heavily pretreated for advanced NSCLC. Patients were subdivided in 3 cohorts: EGFR-mutated, KRAS-mutated, no EGFR- or KRAS-mutated. In this last cohort, 4 partial responses were reported, and all in ALK-positive crizotinib-naïve patients. However, ALK gene rearrangements were retrospectively detected by FISH in 1 case or PCR-based assays in the other 3 cases. The most common adverse events reported in all patients were diarrhoea, fatigue, nausea and anorexia (55). Based on these earlier results, to evaluate whether or not this approach will be effective the CHIARA trial (CHaperone Inhibition in ALK Rearranged lung cAncer) is ongoing. In this phase II trial, ganetespib monotherapy is administered to previously treated patients with stage IIIB/IV NSCLC harbouring an ALK gene rearrangement and who have not been previously treated with a direct ALK inhibitor. Primary endpoint is objective response rate with approximately 100 patients planned to be enrolled (58). To mitigate the development of acquired resistance, a dual inhibition of critical signalling pathways is being investigated. A phase I/II study is evaluating the combination of crizotinib and ganetespib in previously treated patients with NSCLC ALK-positive not pretreated with any specific inhibitor, with the primary endpoint to define the maximum tolerated dose to be investigated in the subsequent phase II trial (59). This trial should give more information on whether this approach is more effective than either therapy alone, and whether dual treatment prevents development of acquired resistance.

Another HSP90 inhibitor is retaspimycin hydrochloride (IPI-504), which was investigated in a phase II trial at the starting dose of 400 mg/m² on days 1, 4, 8 and 11 of a 21-day cycle and then at the dose of 225 mg/m² due to hepatotoxicities observed at the highest dose in another trial. A total of 76 patients with advanced NSCLC and heavily pretreated including a line with EGFR-TKI were enrolled. Among these patients, 3 were ALK-positive, two had partial responses and the third had prolonged stable disease (7.2 months, 24% reduction in tumour size). The most common grades 1 and 2 adverse events included fatigue, nausea and diarrhoea. Grade ≥3 hepatotoxicities were observed in nine patients (11.8%) (56).

AUY922, is a highly potent, non-geldanamycin, HSP90 inhibitor which was tested in a phase II study including also ALK-positive NSCLC patients. AUY922 was administered at the weekly dose of 70 mg/m² to 121 previously treated NSCLC patients. In the total of 22 ALK-positive patients, 7 objective responses (32%) were reported, 3 of which among the 14 crizotinib-resistant NSCLC patients; the disease control rate was 59% (100% in the crizotinib-naive group and 36% in the crizotinib-resistant group). The most frequent adverse events, mainly grade 1–2, were eye disorders (77%), diarrhoea (74%) and nausea (46%) (57). AUY922 is currently being investigated in ALK-positive patients in two trials. A phase II study is evaluating AUY922 in ALK-positive patients resistant to an ALK-TKI therapy with the objective response rate as primary endpoint and the estimated enrolment of 20 patients (60). LDK378 and AUY922 were administered within a phase Ib study to patients with ALK-positive NSCLC already pretreated with crizotinib. The primary endpoint is the incidence rate of dose limiting toxicities with the estimated enrolment of 142 patients (61).

In a retrospective analysis, the efficacy of pemetrexed in terms of PFS was evaluated in 89 advanced NSCLC studying the relationship with specific molecular subtype. Patients with ALK gene rearrangements had a longer median PFS on pemetrexed than on KRAS mutant, EGFR mutant, or triple-negative patients (9 months compared with 7, 5.5 and 4 months, respectively). The data were confirmed also by the multivariate analysis in which the only statistically significant variable associated with prolonged PFS on pemetrexed was ALK-positivity (HR 0.36; p=0.0051) (62). Another retrospective analysis confirmed the previous data. Fifteen ALK-positive patients reported a better objective response rate than 43 EGFR-mutated patients and 37 wild-type patients with 46.7, 4.7 and 16.2% (p=0.001), respectively. Time to progression (TTP) was also in favour of ALK-positive patients (9.2, 1.4 and 2.9 months, respectively; p=0.001) (63).

This correlation might be explained with the lower concentration of thymidylate synthase (TS), the main target of pemetrexed, in ALK-positive tumours (63,64).

The largest retrospective analysis compared 121 ALK-positive NSCLC patients with 266 patients with ALK-negative, EGFR-wild-type NSCLC, including 79 with KRAS mutations. Among 70 ALK-positive patients treated with a platinum/pemetrexed regimen, the median PFS was 7.3 months while it was 5.5 months for 51 ALK-positive patients treated with single-agent pemetrexed or non-platinum/pemetrexed combinations. PFS of ALK-negative patients, on all pemetrexed-based regimens, was similar to that of ALK-positive patients, except in first-line platinum/pemetrexed where the median PFS was 4.2 and 5.4 months, in patients with KRAS-mutation and wild-type, respectively (65).

Contrary to previous data, this large retrospective analysis did not report any statistical advantage for ALK-positive patients treated with pemetrexed-based therapy.

These contrasting results could be directly addressed by the prospective studies such as PROFILE 1007 and PROFILE 1014. To date, only the results of PROFILE 1007 are available, showing an interesting relationship between ALK-positivity and sensitivity to pemetrexed treatment. Patients treated with pemetrexed experienced better objective response rates as compared to those treated with docetaxel (29.3 and 6.9%, respectively) and better PFS (4.2 and 2.6 months, respectively). It is important to emphasise that the study was not designed to compare the two chemotherapeutics. Moreover, the choice between pemetrexed or docetaxel was left to the investigator and not to randomization (35) (Table IV).

The final results of the PROFILE 1014 trial are awaited also to better understand the activity of pemetrexed in ALK-positive NSCLC patients.

ROS1 is a receptor tyrosine kinase of the insulin receptor family, shown to fuse at exons 32, 34, 35 or 36 with multiple partners in NSCLC (TPM3, SDC4, SLC34A2, CD74, EZR, LRIG3 and FIG). These chimeric proteins, found in approximately 1–2% of NSCLC patients, maintain constitutive ROS1 kinase activity, leading to persistent downstream signalling and transforming ability via enhanced cell growth, proliferation and decreased apoptosis (66,67). Like ALK-positive NSCLC patients, ROS1-positive NSCLC patients tend to be younger (median age, 49.5 years), never-smokers, have a histological diagnosis of adenocarcinoma, and seem to be mutually exclusive with other gene alterations (68).

Preclinical studies of cell lines harbouring ROS1 rearrangements and clinical evidence on a single CD74-ROS1-positive NSCLC patient showed sensitivity to crizotinib (68,69). Thirteen ROS1-positive NSCLC patients, all previously treated, received crizotinib at the standard dose of 250 mg twice daily. Patients were young with a median age of 47 years, all with adenocarcinoma histology, and all but one were never-smokers. The objective response rate was 54%, with a disease control rate at 8 weeks of 85% (70). A progression of disease is reported also for ROS1-positive patients after a response to crizotinib treatment. A crizotinib resistance due to an acquired mutation leading to a glycine-to-arginine substitution at codon 2032 in the ROS1 kinase domain was demonstrated in a ROS1-positive patient initially responding to crizotinib. This mutation confers resistance to crizotinib through steric interference with drug binding (71).

The incidence of brain metastases in ALK-positive NSCLC patients was described to be ≤46% (72). A case report showed development of brain metastases under crizotinib therapy, despite an objective response of extracranial lesions, in an ALK-positive NSCLC patient. Measurement of the levels of crizotinib in the plasma and cerebrospinal fluid (CSF) indicated a poor penetration of the drug into the brain, thus potentially hindering efficacy in metastatic brain tumours (73). Further reports underlined this aspect reporting several cases in which ALK-positive NSCLC patients reported a brain progression of disease despite a shrinkage of extracranial tumour sites (74–76). On the contrary, other reports showed slight reduction of brain metastases (77) or response (78) with crizotinib therapy even if with a different schedule.

Interesting results on brain metastases were gained from phase I/II trials with alectinib (50,79). Among the 46 patients enrolled in the phase II portion of the first study (50), 15 (33%) patients with known identified brain metastases, of whom 12 (26%) were previously irradiated and three (7%) were clinically stable without symptoms at baseline. No progression of brain metastases in any of the patients was reported, although previous radiotherapy might have affected the natural history of brain disease. Two of the three patients who had baseline brain metastases and had not received previous radiotherapy continued alectinib for >300 days without brain progression (50). In the second trial (79), 21 out of 47 enrolled patients had brain metastases at baseline, and 4 received no prior brain irradiation. In these 4 patients, two complete and one partial responses with one stable disease were reported. Among the 17 patients with brain metastases and previously irradiated, one patient progressed in both systemic and brain sites, three progressed only in systemic sites and the other 13 patients are still on treatment with one complete response (79).

In both trials, alectinib demonstrated consistent and rapid clinical activity in brain metastases in ALK-positive NSCLC patients who progressed on crizotinib. This is likely due in part to low penetration of crizotinib into the central nervous system. In fact, the blood-brain barrier (BBB) contributes to brain homeostasis by protecting the brain from potentially harmful endogenous and exogenous substances. The efflux transporter P-glycoprotein (Pgp) is a key element of the BBB that can actively transport a huge variety of lipophilic drugs out of the brain capillary endothelial cells that form the BBB (80). Crizotinib is a good Pgp substrate, but not alectinib. Moreover, preclinical studies in the central nervous system implantation models suggest a promising antitumour activity of alectinib against brain lesions (79).

Of course, further studies specifically addressing this issue must better clarify the potential role of alectinib also in this setting.