Systematic review of sequencing of ALK inhibitors in ALK-positive non-small-cell lung cancer.

The objective of this study was to understand outcomes of patients treated with ALK inhibitors, especially when ALK inhibitors are followed by other ALK inhibitors. A systematic literature review was conducted in PubMed, Embase, and Cochrane through July 17, 2017. Conference abstracts (three meetings in past 2 years) also were searched. Of 504 unique publications, 80 met inclusion criteria (47 clinical trials, 33 observational studies). Observational studies have the potential to provide information for ALK inhibitors used sequentially. Ten observational studies reported median overall survival of crizotinib-led sequences ranging from 30.3 to 63.75 months from initiation of crizotinib; 49.4-89.6 months from metastatic non-small-cell lung cancer diagnosis; and 15.5-22.0 months from initiation of the second-generation ALK inhibitor after initial crizotinib. Sequencing of ALK inhibitors may benefit patients progressing on initial ALK inhibitors.

Introduction

ALK is a member of the insulin receptor superfamily,1 and oncogenic EML4-ALK fusion variants represent molecular targets in non-small-cell lung cancer (NSCLC). ALK fusions have been identified in both squamous and adenocarcinoma histologic subtypes, with a higher frequency observed in adenocarcinoma.2,3 Overall, ALK fusions occur in 3%–5% of patients with metastatic NSCLC.4

Prior to 2011, when the first ALK tyrosine kinase inhibitor was approved, the standard of care for patients with ALK-positive NSCLC was chemotherapy, and outcomes were poor, with median overall survival (OS) of ~12 months.5,6 Crizotinib was approved by the United States Food and Drug Administration (FDA) under accelerated approval in 2011 and was the first ALK inhibitor approved for patients with ALK-positive advanced NSCLC.7

Although patients with ALK-positive advanced NSCLC initially respond to ALK inhibitors, resistance eventually often develops in these patients.8 One of the mechanisms of acquired resistance is a mutation in the kinase domain of ALK, although other resistance mechanisms have also been reported, such as activation of alternative pathways (EGFR, KIT, and IGF-IR), ALK amplification, and epithelial–mesenchymal transition.9 In some patients, the mechanism of acquired resistance remains unknown.9

To address resistance, additional ALK inhibitors have been introduced. Ceritinib was approved by the FDA in April 201410 for the treatment of patients with ALK-positive metastatic NSCLC who have progressed on or are intolerant to crizotinib, and in May 2017 it received approval for expanded use to include first-line treatment.11 Subsequently, alectinib received FDA approval in December 2015 for the treatment of patients with ALK-positive metastatic NSCLC who have progressed on or are intolerant to crizotinib12,13 and in November 2017 for first-line treatment.14 Brigatinib received FDA approval in April 2017 for the treatment of patients with ALK-positive metastatic NSCLC who have progressed on or are intolerant to crizotinib.15

The current standard of care for treating ALK-positive NSCLC is the use of ALK inhibitors. Multiple available ALK inhibitors allow the possibility of sequencing these agents to extend patient benefit and improve outcomes. The available ALK inhibitors have different potencies, differential penetration into the central nervous system, unique safety profiles, and different “spectrums” of activity against particular acquired resistance mutations.

Outcomes of ALK inhibitors are well documented in controlled clinical trials; however, less is known about the outcomes associated with sequencing. We hypothesized that sequencing of ALK inhibitors will benefit survival outcomes of patients. Herein, we report the first systematic literature review with an aim to understand the outcomes of patients treated with ALK inhibitors, especially when an ALK inhibitor is followed by another ALK inhibitor.

Material and methods

Electronic literature searches were conducted in PubMed, Embase, and the Cochrane Library databases through July 17, 2017 for real-world and clinical trial evidence for drug sequencing/treatment patterns and the related outcomes associated with the use of ALK inhibitors. Additional studies not published in the peer-reviewed literature were identified by searching online conference abstracts of three professional societies for the previous 2 calendar years: the American Society of Clinical Oncology (2016 and 2017), the European Society of Medical Oncology (2015 and 2016), and the International Association for the Study of Lung Cancer World Conference on Lung Cancer (2015 and 2016). The electronic database searches were also supplemented by a review of the bibliographic reference lists of relevant literature review articles.

The search terms for the medical library databases included Medical Subject Heading, Emtree, and free-text terms, including disease terms (carcinoma, non-small-cell lung; non-small-cell lung cancer; non-small-cell lung carcinoma; non-small-cell lung cancer), terms to identify drug sequencing/treatment patterns (practice pattern, prescribing pattern, treatment pattern), terms to identify the agents of interest (crizotinib, Xalkori, PF-02341066, ceritinib, Zykadia, LDK378, alectinib, Alecensa, CH5424802, brigatinib, AP26113, ALK inhibitor), various terms to identify study types and outcomes of interest, and terms to identify observational studies and clinical trials (Table S1). The search was limited to English-language studies of humans and had no date limit.

Two independent reviewers screened the titles and abstracts according to predefined inclusion and exclusion criteria (Table S2). Full-text articles of selected records were obtained, and the two independent reviewers further screened each article according to the same predefined inclusion and exclusion criteria. Data extraction by a single researcher included study design, patient characteristics, line/sequence of therapy, and outcomes, including treatment duration, response rates, median OS, and median progression-free survival (PFS). A separate researcher conducted quality control of data extraction.

Results

The electronic literature database search identified 481 unique records. One additional article was identified following a review of the bibliographic reference lists of relevant literature review articles. Twenty-two additional abstracts were identified from the search of professional societies and associated conferences. Of the 504 unique articles/abstracts identified, 80 publications met the inclusion criteria (Figure 1). Of the 80 publications, 47 were from clinical trials and 33 were from observational studies. Studies were heterogeneous regarding study design, data source, sample size, timeframe of observation, and outcomes collected, including PFS and OS. A detailed overview of the PFS and OS outcomes in the observational studies of ALK inhibitors used after an initial ALK inhibitor is shown in Tables 1 and 2, respectively. The online supplement provides a list of the 80 publications included (Table S3).

Figure 1

PRISMA diagram.

Table 1

Results of observational studies of an ALK inhibitor after initial ALK inhibitor – median PFS

Reference	Median PFS of first ALK inhibitor, months	Median PFS of second ALK inhibitor, months	Median combined PFS, monthsa
Ceritinib after initial crizotinib
Bendaly et al81	NR	NR	NR
Bendaly et al80	NR	12.9	NR
Gainor et al82	8.2	7.8	17.4 17.0 for patients with no interval between CRZ and CERT
Kayaniyil et al48	NR	9.6 for patients who received CERT immediately after CRZ 4.6 for patients who received CERT at any time after CRZ	NR
Alectinib after initial crizotinib
Ito et al44	7.0	24.7b	NR
Watanabe et al84	6.1	15.2c	18.2
Asao et al46	10.7	16.6d	35.2
Yoshida et al57	NR	NR	NR
Any second-generation ALK inhibitor after crizotinib
Chiari et al45	10	7	17
Roeper et al83	NR	NR	NR
Nosaki et al47	NR	NR	NR
Cadranel et al50	NR	NR	NR
Duruisseaux et al54	NR	NR	NR
Second-generation ALK inhibitor after second generation
Nosaki et al47	NR	NR	NR

Notes:

Summarized as reported in the publication, ie, the authors of this review did not calculate the results.

Retrospective study of 28 patients in whom 15 received alectinib alone and 13 received alectinib after initial crizotinib.

Retrospective study of 11 patients; all patients received crizotinib followed by alectinib but 6 of the 11 also received ≥1 cytotoxic chemotherapy regimen prior to crizotinib. Range of median PFS on alectinib was 1.0–28.3.

Retrospective study of 13 patients who received crizotinib followed by alectinib.

Abbreviations: CERT, ceritinib; CRZ, crizotinib; NR, not reported; PFS, progression-free survival.

Table 2

Results of observational studies of an ALK inhibitor after initial ALK inhibitor – median OS

Reference	Median OS from initiation of second ALK inhibitor, months	Median OS from initiation of first ALK inhibitor (ie, OS of ALK sequence), months	Median OS from diagnosis of metastatic disease, months
Ceritinib after initial crizotinib
Bendaly et al81	NR	NR	NR
Bendaly et al80	15.5	NR	NR
Gainor et al82	NR	30.3a	49.4
Kayaniyil et al48	20.4	NR	51.0
Alectinib after initial crizotinib
Ito et al44	NR	Not reachedb	NR
Watanabe et al84	NR	48.6	51.1
Asao et al46	NR	NR	NR
Yoshida et al57	NR	NR	NR
Any second-generation ALK inhibitor after crizotinib
Chiari et al45	22	40	NR
Roeper et al83	NR	Not reachedc	NR
Nosaki et al47	NR	63.75	NR
Cadranel et al50	NR	Not reachedd	NR
Duruisseaux et al54	NR	Not reachede	89.6
Second-generation ALK inhibitor after second generation
Nosaki et al47	NR	Not reachedf	NR

Notes:

From initiation of crizotinib in subset of patients previously treated with chemotherapy.

Median duration of follow-up: 21.3 months.

Median duration of follow-up: not reported.

Median duration of follow-up: 21.4 months.

Median duration of follow-up: 44.4 months.

Median duration of follow-up: not reported.

Abbreviations: NR, not reported; OS, overall survival.

Evidence base of first use of an ALK inhibitor with or without prior chemotherapy (ALK inhibitor naïve)

A total of 45 publications assessed outcomes of first use of an ALK inhibitor with or without prior chemotherapy in patients who were ALK inhibitor naïve. Of the 45 publications, 27 were from clinical trials16–42 and 18 were from observational studies.43–60 In clinical trials, median PFS ranged from 7.7 months30 to 25.9 months,19 median OS ranged from 20.3 months30 to 39.1 months,20 and objective response rate (ORR) ranged from 46%27 to 100%.36 In the observational studies, median PFS ranged from ~7 months (reported as 28 weeks)51 to 17.7 months (from diagnosis of advanced NSCLC)58 and median OS ranged from 11.2 months54 to ~104 months (reported as 416 weeks);47 note that data from Nosaki et al47 were presented in a conference abstract and thus not all data may have been included.

Evidence base of use of second or subsequent ALK inhibitor (ALK inhibitor followed by another ALK inhibitor)

A total of 38 publications assessed outcomes of use of an ALK inhibitor after an initial ALK inhibitor. Of the 38 publications, 25 were from clinical trials35–37,40–42,61–79 and 13 were from observational studies.44–48,50,54,57,80–84 All 38 publications reported on use of a second-generation ALK inhibitor after initial crizotinib therapy; one publication of an observational study also included a population that used a second-generation ALK inhibitor after an initial second-generation ALK inhibitor (ceritinib after initial alectinib),47 and one publication of an observational study mentioned two patients who received crizotinib after alectinib.84 The efficacy data reported in clinical trials are from initiation of the second-generation ALK inhibitor only; as noted, all sequences were of a second-generation ALK inhibitor after initial crizotinib therapy. In the 25 publications from clinical trials, median PFS from initiation of the second-generation ALK inhibitor ranged from 5.4 months77 to 15.6 months,70 median OS ranged from 14.9 months75 to 26.0 months,61,65 and ORR ranged from 33.0%42 to 80.0%;36 estimated 12-month survival rates ranged from 63.8%75 to 83.0%.36 Only seven clinical trial publications reported treatment duration of the second-generation ALK inhibitor, which ranged from 8.1 weeks76 to 38.6 weeks.37

Of the 13 publications from observational studies, there was a heterogeneous makeup in study design, population, and the method in which results were reported. Combined PFS of a crizotinib-led sequence was reported in four publications45,46,82,84 and ranged from 17.0 months45,82 to 35.2 months.46 In general, median combined PFS was defined as the sum of PFS of the two ALK inhibitors and did not include the interval from discontinuation of crizotinib to initiation of the second-generation ALK inhibitor or postprogression use of crizotinib. In some instances, patients may have been allowed to receive chemotherapy in the interval between use of the two ALK inhibitors. Five observational study publications44–46,82,84 of crizotinib-led sequences reported PFS of crizotinib ranging from 6.1 months84 to 10.7 months.46 Seven publications44–46,48,80,82,84 reported PFS from initiation of the second-generation ALK inhibitor ranging from 4.6 months48 to 24.7 months.44

Ten publications from observational studies44,45,47,48, 50,54,80,82–84 of crizotinib-led sequences reported median OS. Of these ten publications, four reported median OS for the sequence as not reached.44,50,54,83 Median follow-up for the sequence was reported as follows: not reported,83 21.3 months,44 21.4 months,50 and 44.4 months.54 Four publications45,47,82,84 reported median OS from initiation of crizotinib (ie, for the full ALK sequence) ranging from 30.3 months82 to ~64 months (reported as 255 weeks).47 The 64 months reported by Nosaki et al47 was in patients receiving alectinib or ceritinib after initial crizotinib; however, this publication was a conference abstract and thus limited data are reported. Four publications48,54,82,84 reported median OS from diagnosis of metastatic disease ranging from 49.4 months82 to 89.6 months.54 The 89.6 months reported by Duruisseaux et al54 were in 84 patients who received a second-generation ALK inhibitor at some point after progressing on initial crizotinib. Three publications45,48,80 reported median OS of the second-generation ALK inhibitor after crizotinib ranging from 15.5 months80 to 22.0 months.45 One publication of a second-generation ALK inhibitor after an initial second-generation ALK inhibitor (ceritinib after initial alectinib) reported median OS for the sequence as not reached; median duration of follow-up was not reported.47

Survival rates were reported in five publications from observational studies.46,48,50,54,80 Twelve-month survival rates in crizotinib-led sequences were reported as 59.9%50 and 92.9%54 from first dose of crizotinib.

Specific sequences

Ceritinib after initial crizotinib

When reviewing specific sequences, 12 publications described results of ceritinib after initial crizotinib; 8 were from clinical trials,37,40–42,75–77,79 and 4 were from observational studies.48,80–82 The clinical trials reported results for only the ceritinib portion of the sequence. Median PFS ranged from 5.4 months77 to 6.9 months,37,41 median OS ranged from 14.9 months75 to 20.0 months,79 and ORR ranged from 33% to 63%.42

In the four observational study publications, combined median PFS was reported in only one publication and was reported as 17.4 months.82 Combined PFS for sequential treatment with crizotinib and ceritinib did not include postprogression use of crizotinib or the interval between crizotinib discontinuation and start of ceritinib, in which patients could have received cytotoxic chemotherapy.82 In patients in which crizotinib was discontinued and ceritinib immediately initiated (ie, no intervening treatment), median combined PFS was 17.0 months.82 Median PFS while patients were on ceritinib ranged from 7.8 months82 to 12.9 months.80 Only one study reported median PFS while patients were on crizotinib, which was reported as 8.2 months.82

Finally, median OS was reported as 30.3 months from initiation of crizotinib,82 15.5 months80 and 20.4 months48 from initiation of ceritinib, and 49.4 months82 and 51.0 months48 from diagnosis of metastatic disease.

Alectinib after initial crizotinib

A total of 15 publications described results of alectinib after initial crizotinib; 11 were from clinical trials,35,61–69,78 and 4 were from observational studies.44,46,57,84 The clinical trials reported results for just the alectinib portion of the sequence. Median PFS ranged from 8.0 months64 to 13.9 months,78 median OS was reported as 22.7 months64 and 26.0 months,61,65 and ORR ranged from 44.0%62 to 72.2%.35 The estimated 12-month survival rate was 71.0%.63

Of the four observational studies, two publications reported combined median PFS as 18.2 months84 and 35.2 months.46 Asao et al46 defined combined PFS as the sum of the PFS of the two ALK inhibitors without the interval between the ALK inhibitors, ie, treatment duration with cytotoxic chemotherapy between the two ALK inhibitors was excluded. In Watanabe et al,84 combined PFS did not include postprogression use of crizotinib or the interval from discontinuation of crizotinib to initiation of alectinib. Median PFS from initiation of alectinib ranged from 15.2 months84 to 24.7 months.44 Median PFS while patients received crizotinib ranged from 6.1 months84 to 10 months.45

Median OS was reported to be not reached44 (median follow-up of 21.3 months) and 48.6 months84 from initiation of crizotinib and 51.1 months84 from diagnosis of metastatic disease. Estimated 12-month survival was reported to be 38.6% in patients on crizotinib and 60.0% in those on ceritinib;46 the estimated 5-year survival for the sequence was 77.8%.46

Brigatinib after initial crizotinib

Six publications described results of brigatinib after initial crizotinib, all of them clinical trials.36,70–74 There were no observational studies found in the literature that assessed brigatinib after initial crizotinib. Median PFS from initiation of brigatinib ranged from 8.8 months70 to 15.6 months,70 and ORR ranged from 45%70,73 to 80%.36 Median OS was not reported in any of the six publications. Estimated 12-month survival rates ranged from 71%73 to 83%.36

Any second-generation ALK inhibitor after initial crizotinib

Five publications from observational studies described outcomes of a second-generation ALK inhibitor after initial crizotinib.45,47,50,54,83 In these publications, either the second-generation ALK inhibitor was not specified or the results were combined for more than one ALK inhibitor. All five publications were from observational studies. Median PFS was reported in only one publication and was 17 months combined for the sequence, 10 months for crizotinib, and 7 months from initiation of the second ALK inhibitor (alectinib or ceritinib).45 Median OS was 40.0 months45 and ~64 months47 for the sequence, 22 months from initiation of the second ALK inhibitor,45 and 89.6 months from diagnosis of metastatic disease.54 Two publications reported median OS as not reached for the specific sequence being studied54,83 (median follow-up not reported in Roeper et al83 and 44.4 months in Duruisseaux et al54). Estimated 12-month survival was reported to be 59.9% from the start of a sequence of ceritinib, alectinib, or brigatinib after initial crizotinib50 and 92.9% from the start of a sequence of ceritinib or alectinib after initial crizotinib.54

Ceritinib after initial alectinib

Nosaki et al47 reported median OS as not being reached in patients who received ceritinib after initial alectinib; median follow-up time was not reported. Note these data were reported from a conference abstract with limited information.

Discussion

In this systematic literature review, we aimed to understand the outcomes of patients treated with ALK inhibitors, especially when an ALK inhibitor is followed by another ALK inhibitor.

The identified clinical trials of patients who were ALK inhibitor naïve reported median PFS ranging from 7.7 to 25.9 months and median OS ranging from 20.3 to 39.1 months. Observational studies reported median PFS ranging from 7 to 17.7 months and median OS ranging from 11.2 to 104 months.

In clinical trials of a second-generation ALK inhibitor used after initial crizotinib, median PFS from initiation of the second-generation ALK inhibitor ranged from 5.4 to 15.6 months and median OS ranged from 14.9 to 26.0 months. In observational studies of an ALK inhibitor followed by another ALK inhibitor, median PFS ranged from 4.6 to 35.2 months and median OS ranged from 15.5 to 89.6 months.

In sequencing observational studies of an ALK inhibitor used after an initial ALK inhibitor, median OS from initiation of the first ALK inhibitor, ie, for the ALK sequence, has varied and has been reported as 30.3 months,82 40 months,45 48.6 months,84 and ~64 months.47 Median OS has been consistently reported to be ~50 months from time of diagnosis of metastatic disease in several observational studies of ALK inhibitors used in sequence,48,82,84 indicating that sequential use of ALK inhibitors may be clinically beneficial to patients. There are currently several examples of median OS being reported as “not reached” in studies of the full sequence of an ALK inhibitor after an initial ALK inhibitor. This is not surprising given that only relatively recently multiple ALK inhibitors became available. As sequential ALK inhibitors are utilized and survival data mature, we expect that additional outcomes data will become available to help inform treatment decisions for improved outcomes of patients with ALK-positive NSCLC. Important to note is that lorlatinib, a third generation ALK inhibitor, recently became available in the USA. Lorlatinib is indicated for the treatment of patients with ALK-positive metastatic NSCLC whose disease has progressed on crizotinib and at least one other ALK inhibitor; or alectinib or ceritinib as the first ALK inhibitor therapy for metastatic disease,85 which further supports the sequencing of ALK inhibitors. Approval was based on a phase 2 study in which lorlatinib demonstrated substantial overall and intracranial response both in treatment-naive patients with ALK-positive NSCLC, and in those who had progressed on crizotinib, second-generation ALK inhibitors, or after up to three previous ALK inhibitors.86

The evidence base is broader and more mature for crizotinib-led sequences than for second-generation-led sequences. The amount of research of crizotinib-led sequences is not unexpected given that crizotinib was the first ALK inhibitor on the market. Additional research is needed to understand the survival outcomes of second-generation ALK inhibitors as initial therapy.

This study adds to the current literature in that it is the first systematic review of sequencing of ALK inhibitors. A robust methodology was used that included a study protocol, multiple broad electronic databases searching for both clinical trials and observational studies, and that did not limit by date. In addition, two reviewers independently screened all titles, abstracts, and full-text articles using predefined inclusion and exclusion criteria. This robust methodology enables the reproducibility of the review.

It is important to note in the interpretation of retrospective studies reporting median OS or “combined” PFS for a sequence that immortal time bias must be considered. In studies of sequential therapy conducted retrospectively, patients who do not survive to receive the second treatment are not included in the analysis. Patients who received both ALK inhibitors are selected for having lived long enough and for having stable enough disease (in some instances related in part to chemotherapy after the first ALK inhibitor) that they were able to receive both ALK inhibitors in sequence. Therefore, the observed value for combined PFS and OS reported in these studies is likely to be biased upward from what may be expected at the outset for patients treated according to such a sequential treatment plan; however, as no prospectively designed studies have evaluated this question to date, retrospective studies are currently the best available evidence.

Another important consideration in interpreting the findings from this systematic review is that the cutoff date of July 17, 2017 did not allow for inclusion of the final OS data from PROFILE 1014, which is the first long-term study with mature OS data for an ALK-positive NSCLC population. Results showed that median OS was not reached for crizotinib and was 47.5 months for chemotherapy (median follow-up was ~46 months in each treatment arm).87 Most patients (84.2%) receiving chemotherapy crossed over to crizotinib; therefore, a crossover-adjusted analysis was conducted demonstrating OS in the crizotinib arm to be significantly longer than the chemotherapy arm (HR, 0.346; 95% CI, 0.081–0.718).87 At 4 years, 56.6% of crizotinib patients and 49.1% of chemotherapy patients were still alive.87 Interestingly, these results were consistent with the OS data of around 50 months from observational studies of ALK inhibitors that we identified in this systematic literature review.

Owing to the data immaturity and currently available trial designs, it is not currently possible to determine which sequence confers the best long-term outcomes.

Conclusion

Subsequent use of ALK inhibitors may clinically benefit patients progressing on an initial ALK inhibitor. Crizotinib-led sequences have a broader evidence base and more mature clinical outcomes than second-generation-led sequences. No evidence was found directly comparing different ALK inhibitor sequences. Further research is warranted to directly compare ALK inhibitor sequences and to understand the outcomes of second-generation ALK inhibitors as initial ALK inhibitor therapy.