Background
Lung cancer remains the leading cause of cancer-related death worldwide, with non-small-cell lung cancer (NSCLC) accounting for approximately 85% of cases [1]. Targeted therapy plays a pivotal role in the management of NSCLC by targeting actionable genomic alterations that drive tumor growth and proliferation [2,3]. Among these alterations, anaplastic lymphoma kinase (ALK) gene rearrangements represent an important oncogenic driver and actionable therapeutic target in NSCLC [4,5].
Patients with ALK-rearranged NSCLC generally derive clinical benefit from treatment with ALK tyrosine kinase inhibitors [6,7]. First-generation ALK inhibitors, such as crizotinib, represented a milestone in the treatment of ALK-rearranged NSCLC but provided relatively modest disease control, with median progression-free survival (PFS) ranging from 7 to 11 months in patients with previously untreated ALK-positive NSCLC despite relatively high objective response rates (ORRs) of approximately 60% to 74% [8,9]. This limited clinical benefit largely reflects the emergence of acquired resistance, with a limited duration of response (DoR) and disease progression commonly occurring within the first year of therapy [8,9].
These limitations highlight the need for novel ALK inhibitors that provide more durable responses and improve clinical outcomes in patients with ALK-positive NSCLC.
ALK-positive advanced NSCLC carries a substantial lifetime risk of central nervous system (CNS) involvement, with 20% to 30% of patients presenting with brain metastases at diagnosis [10,11] and cumulative incidences of 45.5% and 58.4% at 2 and 3 years after diagnosis, respectively [12]. The presence of CNS metastases in ALK-positive NSCLC is associated with poor overall survival (OS) and diminished quality of life [10]. Brain metastases frequently contribute to treatment failure and disease progression because many systemic therapies have limited penetration across the blood–brain barrier (BBB) [13,14].
First-generation ALK inhibitors have poor CNS penetration, primarily because they are substrates for P-glycoprotein-mediated efflux across the BBB [15,16]. Notably, following crizotinib treatment, progression of brain metastases occurs in up to 50% to 60% of patients [17]. Therefore, novel ALK inhibitors that effectively control brain metastases while maintaining a favorable CNS safety profile are warranted.
Dirozalkib is a novel, next-generation, highly selective ALK inhibitor that has demonstrated potent inhibitory activity against a broad spectrum of resistance-associated ALK mutations, including mutations associated with resistance to first- and second-generation ALK inhibitors, such as C1156Y, L1196M, F1174L, G1202R, and I1171N/T/S. Notably, dirozalkib retains potent activity in cell lines harboring alectinib-resistant mutations and has demonstrated greater in vivo efficacy than ceritinib in crizotinib-resistant patient-derived xenograft models, as well as potent intracranial antitumor activity. A phase I clinical trial (NCT05055232) demonstrated promising antitumor activity and a favorable safety and tolerability profile with dirozalkib in patients with ALK-rearranged advanced NSCLC, particularly among ALK inhibitor-naïve patients [18].
Thus, the present phase III, multicenter, randomized, open-label clinical trial (DIAMOND-2) aimed to evaluate the efficacy and safety of dirozalkib versus crizotinib in patients with ALK inhibitor-naïve, ALK-positive advanced NSCLC.
Materials and Methods
Study design and patients
The DIAMOND-2 trial (NCT05204628) was a phase III, multicenter, randomized, open-label study comparing dirozalkib with crizotinib in ALK inhibitor-naïve patients with ALK-positive advanced NSCLC. The trial was initiated at 64 hospitals in China, of which 52 ultimately enrolled participants. The study was conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice guidelines.
The protocol (Data S1) was approved by the ethics committee of the lead center, Jilin Province Cancer Hospital (202111-106-07), as well as by the ethics committees at each participating site. All patients provided written informed consent before enrollment. An independent data monitoring committee periodically reviewed safety data and study conduct to ensure participant safety and the integrity of the trial.
Eligible participants were adults aged 18 to 75 years with pathologically confirmed advanced NSCLC and ALK-positive status confirmed by immunohistochemistry, fluorescence in situ hybridization, polymerase chain reaction, or next-generation sequencing (NGS). Eligible participants had not previously received an ALK inhibitor and either had not received systemic anticancer therapy or had received one prior chemotherapy regimen but subsequently experienced disease progression or intolerance. Participants were required to have an Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2 and at least one measurable extracranial lesion according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 that had not been previously irradiated.
Patients without brain metastases or with asymptomatic, untreated brain metastases were eligible. Key exclusion criteria included small cell lung cancer, uncontrolled malignant pleural, peritoneal, or pericardial effusion that reaccumulated markedly within 1 week after drainage, or active infections deemed to pose safety concerns. Detailed inclusion and exclusion criteria are provided in the study protocol (Data S1).
Treatment and procedures
Participants were randomly assigned in a 1:1 ratio to receive either dirozalkib (500 mg orally, once daily) or crizotinib (250 mg orally, twice daily) using stratified block randomization through an interactive web response system. The stratification factors were the presence of brain metastases (yes versus no) and prior chemotherapy (yes versus no). Each treatment cycle lasted 28 d, and treatment continued until disease progression, unacceptable toxicity, withdrawal of consent, death, or discontinuation at the discretion of the investigator or patient, whichever occurred first.
Dose adjustments, treatment interruptions, or permanent discontinuation of dirozalkib or crizotinib was permitted in the event of adverse events (AEs).
Radiologic assessments using computed tomography and/or MRI were performed at baseline, every 8 weeks during the first 64 weeks of treatment, and every 12 weeks thereafter. Tumor responses were evaluated according to RECIST version 1.1 and the Response Assessment in Neuro-Oncology Brain Metastases criteria. AEs were monitored from the time of informed consent until at least 30 d after the final dose of study treatment and were graded according to the Common Terminology Criteria for Adverse Events version 5.0.
AE monitoring included assessments of vital signs, physical examinations, echocardiography, electrocardiography, and laboratory parameters.
Outcomes
The primary end point was investigator-assessed PFS, defined as the time from randomization to documented disease progression or death from any cause, whichever occurred first. Secondary end points included PFS assessed by blinded independent central review (BICR), ORR, disease control rate (DCR), DoR, OS, intracranial ORR, intracranial DoR, and safety. The cumulative incidence of intracranial progression (ICP) and investigator-assessed intracranial PFS were evaluated in post hoc analyses.
Statistical analysis
A parallel-group superiority design was used, assuming a median PFS of 11.0 months with crizotinib and 20.0 months with dirozalkib. With a one-sided type I error rate of 0.025 and 95% power, 147 PFS events were required to detect a hazard ratio (HR) of 0.550. Assuming an accrual period of approximately 18.0 months and a follow-up duration of 20.0 months after the last patient was randomized, an interim analysis was planned when 70% of the required PFS events had occurred.
The α-spending function for the interim analysis followed the O’Brien–Fleming approach. Based on these assumptions, a total of 201 participants were required. To account for a potential dropout rate of 15%, the target enrollment was increased to at least 238 patients (119 per group).
Because the study was conducted during the COVID-19 pandemic, the planned enrollment was increased to approximately 268 participants to accommodate an expected dropout rate of 25%, while the required number of PFS events remained unchanged. We report the final analysis of PFS, which was conducted after the prespecified number of PFS events had occurred.
Efficacy analyses were conducted in the intention-to-treat (ITT) population, which included all randomized participants. Additional post hoc analyses were also performed in the ITT population. The modified ITT (mITT) population, used for a post hoc efficacy analysis, comprised participants in the ITT population for whom information about the regulatory approval status of the NGS kits used for baseline ALK rearrangement testing was available.
Safety analyses were performed in the as-treated population, which included all participants who received at least one dose of study treatment.
Time-to-event end points were estimated using the Kaplan–Meier method, and between-group comparisons were performed using one-sided stratified log-rank tests. HRs and corresponding 95% confidence intervals (CIs) were estimated using Cox proportional hazards regression models. The Cochran–Mantel–Haenszel test was used to compare ORR and DCR between treatment groups, and corresponding 95% CIs were calculated for each treatment group.
All statistical analyses were performed using SAS, version 9.4 or later (SAS Institute, Cary, NC, USA). Two-sided P < 0.05 was considered statistically significant.
Results
Patients and disposition
Between 2022 January 24 and 2023 April 30, a total of 275 patients at 52 hospitals were enrolled and randomly assigned to receive either dirozalkib (n = 137) or crizotinib (n = 138) (Fig. 1). Baseline demographic and clinical characteristics were generally comparable between the 2 groups, including brain metastases (21.2% versus 21.0%), prior brain radiotherapy (1.5% versus 1.4%), and prior chemotherapy for advanced disease (8.8% versus 10.9%; Table 1).
At the data cutoff (2025 May 31), 76 patients (55.5%) in the dirozalkib group and 110 patients (79.7%) in the crizotinib group had discontinued treatment, whereas 61 and 28 patients, respectively, remained on treatment.
Efficacy among the ITT population
As of 2025 May 31, the median follow-up durations in the ITT population were 32.8 (range, 2.2 to 39.0) months in the dirozalkib group and 31.6 (range, 0.3 to 39.5) months in the crizotinib group. A total of 158 PFS events had occurred, including 63 events (46.0%) in the dirozalkib group and 95 events (68.8%) in the crizotinib group. As shown in Fig.
2A, the median investigator-assessed PFS was 29.5 (95% CI = 20.3 to not reached [NR]) months in the dirozalkib group and 12.9 (95% CI = 11.1 to 17.6) months in the crizotinib group (HR = 0.47, 95% CI = 0.34 to 0.66; P < 0.001). The PFS benefit with dirozalkib was observed in patients both with and without baseline brain metastases. Among patients with baseline brain metastases, median PFS was significantly longer with dirozalkib than with crizotinib (18.5 versus 9.2 months; HR = 0.38, 95% CI = 0.20 to 0.72; P = 0.001).
Among patients without baseline brain metastases, median PFS was not reached (95% CI = 25.8 to NR) with dirozalkib versus 17.6 (95% CI = 12.9 to 22.9) months with crizotinib (HR = 0.50, 95% CI = 0.34 to 0.73; P < 0.001). PFS benefits favoring dirozalkib over crizotinib were generally observed across prespecified subgroups (Fig. 2B).
BICR-assessed PFS in the ITT population was also significantly longer with dirozalkib than with crizotinib (median PFS, 27.5 versus 12.9 months; HR = 0.54, 95% CI = 0.39 to 0.74; P < 0.001) (Fig. 2C).
As shown in Table 2, the investigator-assessed ORR was 89.1% (95% CI = 82.6% to 93.7%) in the dirozalkib group and 81.9% (95% CI = 74.4% to 87.9%) in the crizotinib group. The median DoR was 29.5 (95% CI = 19.4 to NR) months with dirozalkib and 11.1 (95% CI = 9.3 to 18.5) months with crizotinib (HR = 0.45, 95% CI = 0.32 to 0.64; P < 0.001). The BICR-assessed ORR was 89.8% (95% CI = 83.5% to 94.3%) in the dirozalkib group and 89.1% (95% CI = 82.7% to 93.8%) in the crizotinib group.
The BICR-assessed median DoR was 25.7 months with dirozalkib and 12.9 months with crizotinib (HR = 0.52, 95% CI = 0.37 to 0.74; P < 0.001).
Among patients with measurable brain metastases at baseline, the investigator-assessed intracranial ORR was significantly higher in the dirozalkib group than in the crizotinib group (92.3% versus 11.1%; P < 0.001). The median intracranial DoR was also markedly longer with dirozalkib than with crizotinib (18.4 versus 3.6 months; P < 0.001) (Table S1). In a post hoc analysis of investigator-assessed intracranial PFS in the ITT population regardless of baseline brain metastasis status, intracranial PFS was significantly longer with dirozalkib than with crizotinib (median, NR versus 29.2 months; HR = 0.47, 95% CI = 0.30 to 0.74; P < 0.001), and the 2-year cumulative incidence of ICP was also significantly lower with dirozalkib than with crizotinib (26.1% [95% CI = 18.8% to 35.7%] versus 42.5% [95% CI = 33.3% to 53.2%]; P = 0.001).
The investigator-assessed median intracranial PFS for patients with any brain metastases was 20.3 months (95% CI = 12.9 to NR) in the dirozalkib group and 10.8 months (95% CI = 7.9 to 12.8) in the crizotinib group (HR = 0.42, 95% CI = 0.21 to 0.84; P = 0.006).
At the data cutoff, OS data remained immature, with 66 deaths reported: 32 (23.4%) in the dirozalkib group and 34 (24.6%) in the crizotinib group. Information about subsequent anticancer therapies administered after disease progression is summarized in Table S2. Among patients who discontinued study treatment after disease progression, 54 of 63 patients (85.7%) in the dirozalkib group and 83 of 92 patients (90.2%) in the crizotinib group received subsequent systemic therapy.
In the crizotinib group, 82.6% (76 of 92) of patients received further ALK-targeted therapy, whereas the corresponding proportion was 77.8% (49/63) in the dirozalkib group.
Efficacy among the mITT population
In the post hoc analysis of the mITT population, 131 patients in the dirozalkib group and 136 patients in the crizotinib group were included. As of 2025 May 31, 152 PFS events had occurred in the mITT population. The median PFS was 31.3 (95% CI = 22.9 to NR) months with dirozalkib and 12.9 (95% CI = 11.1 to 17.6) months with crizotinib (HR = 0.47, 95% CI = 0.33 to 0.65; P < 0.001).
The investigator-assessed ORR was 88.5% in the dirozalkib group and 81.6% in the crizotinib group. The median DoR was 32.1 months with dirozalkib and 11.1 months with crizotinib (HR = 0.44, 95% CI = 0.31 to 0.63; P < 0.001).
Among patients with measurable brain metastases at baseline, the intracranial ORR was 91.7% in the dirozalkib group and 11.1% in the crizotinib group. The median intracranial DoR was NR with dirozalkib versus 3.6 months with crizotinib (HR = 0.00, 95% CI = 0.00 to NR; P = 0.001). In the mITT population, the median intracranial PFS was NR with dirozalkib versus 29.2 months with crizotinib (HR = 0.45, 95% CI = 0.28 to 0.71; P < 0.001).
Safety
During the treatment period, the median durations of exposure to dirozalkib and crizotinib were 715 (range, 4 to 1,187) d and 391 (range, 3 to 1,200) d, respectively, with corresponding median relative dose intensities of 99.6% and 98.3%. Treatment-emergent AEs (TEAEs) were reported in all patients in both groups. The most frequently observed TEAEs (incidence of ≥10% in either group and an absolute difference of ≥5% between the 2 groups) are summarized in Table 3.
The most common TEAEs among patients receiving dirozalkib were diarrhea (97.8%), vomiting (88.3%), nausea (67.2%), increased alanine aminotransferase (62.0%), increased aspartate aminotransferase (59.1%), increased blood creatine kinase-MB (46.7%; the MB isoform, composed of M-type and B-type subunits), and weight loss (46.0%), most of which were grade 1 or 2. Grade ≥3 TEAEs occurred in 60.6% of patients in the dirozalkib group and 72.5% of those in the crizotinib group. The most common grade ≥3 TEAEs in the dirozalkib group were diarrhea (19.0%), increased γ-glutamyl transferase (9.5%), hypertriglyceridemia (7.3%), hyperglycemia (6.6%), increased alanine aminotransferase (5.8%), QT interval prolongation (5.8%), and hypokalemia (5.8%) (Table 3).
Treatment-emergent serious AEs were reported in 21.2% of patients treated with dirozalkib and 26.8% of those treated with crizotinib. The incidence of treatment-related serious AEs was similar between the 2 groups (13.9% versus 12.3%) (Table S3).
TEAEs led to dose interruptions in 44.5% of patients in the dirozalkib group and 56.5% of those in the crizotinib group (Table S4). COVID-19 infection was the most common TEAE leading to treatment interruption, affecting 16 patients (11.7%) in the dirozalkib group and 17 patients (12.3%) in the crizotinib group. Dose reductions due to TEAEs occurred in 19.7% of dirozalkib-treated patients and 30.4% of crizotinib-treated patients.
The most common TEAEs leading to dose reductions in the dirozalkib group were increased alanine aminotransferase (4.4%), diarrhea (3.6%), vomiting (3.6%), increased aspartate aminotransferase (2.9%), and hyperglycemia (2.9%). Discontinuations due to TEAEs occurred in 2.2% of patients in the dirozalkib group and 7.2% of those in the crizotinib group, with treatment-related discontinuations occurring in 1.5% and 5.1% of patients, respectively. Five patients died because of TEAEs: 1 patient (0.7%) in the dirozalkib group (cardiac arrest) and 4 patients (2.9%) in the crizotinib group (cardiac arrest and respiratory arrest [n = 1], pneumonia [n = 2], and respiratory failure [n = 1]).
Of note, the cardiac arrest and respiratory arrest events in the crizotinib group occurred in the same patient. None of the fatal TEAEs were considered treatment-related.
Discussion
In this randomized phase III study, the novel ALK inhibitor dirozalkib significantly prolonged PFS compared with crizotinib, corresponding to a 53% reduction in the risk of disease progression or death. The ORR was comparable between the dirozalkib and crizotinib groups, whereas the median DoR was significantly longer with dirozalkib than with crizotinib (29.5 versus 11.1 months). Dirozalkib demonstrated superior efficacy to crizotinib in controlling intracranial disease, with a significantly higher intracranial response rate and more durable intracranial responses.
Grade ≥3 TEAEs, serious AEs, and TEAEs leading to treatment discontinuation were less frequent with dirozalkib than with crizotinib.
Previous studies have demonstrated that second-generation ALK inhibitors, including alectinib, brigatinib, and ensartinib, markedly improve PFS compared with the first-generation ALK inhibitor crizotinib in treatment-naïve patients with advanced ALK-positive NSCLC, with median PFS ranging from 24.0 to 34.8 months and HRs ranging from 0.47 to 0.51 [19–22]. In the present study, the magnitude of the PFS benefit with dirozalkib versus crizotinib (HR = 0.47) was numerically comparable to that reported for these next-generation ALK inhibitors. However, no head-to-head comparisons with second- or third-generation ALK inhibitors were conducted; therefore, the relative efficacy of dirozalkib within the current treatment landscape remains to be established.
The strong agreement between investigator and BICR assessments further underscores the reliability of these findings. Beyond PFS, although the ORR was only modestly higher with dirozalkib than with crizotinib (89.1% versus 81.9%), dirozalkib induced more durable systemic responses, with a median DoR of 29.5 months versus 11.1 months. This finding may be consistent with the potent pharmacodynamic activity of dirozalkib against multiple resistance-associated ALK mutations following crizotinib therapy.
Although the survival data remain immature, dirozalkib appears to be a valuable first-line treatment option for patients with advanced ALK-positive NSCLC.
Among patients with measurable brain metastases at baseline, dirozalkib achieved a markedly higher intracranial ORR than crizotinib (92.3% versus 11.1%), while the advantages in the cumulative incidence of ICP and intracranial PFS further underscored the superior CNS disease control with dirozalkib compared with crizotinib. This benefit likely reflects the optimized CNS pharmacologic profile of dirozalkib, including enhanced BBB penetration and sustained activity against intracranial tumor clones that exhibit resistance to first-generation ALK inhibition. Supporting these clinical findings, preclinical evaluation in the H2228-luc human lung cancer intracranial xenograft model demonstrated that dirozalkib markedly inhibited intracranial tumor growth and prolonged survival compared with vehicle controls [18].
Notably, brain metastases were identified in approximately 21.1% of patients at baseline, whereas prior brain radiotherapy was reported in approximately 1.5% of patients in the present study. These characteristics were similar to those reported in previous studies of second-generation ALK inhibitors, in which baseline brain metastases were documented in approximately 26% to 42% of patients, although the proportion of patients with prior radiotherapy in the present study was lower than the previously reported range of 4.9% to 40% [9,11,23]. Considering the association of prior brain radiotherapy with improved intracranial disease control and survival outcomes [24,25], potential confounding due to differences in patient characteristics should be acknowledged when comparing CNS outcomes across studies.
Collectively, these results demonstrate the potent intracranial efficacy of dirozalkib and reinforce the relevance of CNS penetration as a distinguishing therapeutic advantage over crizotinib.
The safety profile of dirozalkib in this study was consistent with that reported in the prior phase I study [18], with no new safety concerns identified. Although patients receiving dirozalkib had a longer median treatment duration than those receiving crizotinib, grade ≥3 TEAEs, serious AEs, and TEAEs resulting in dose interruption or reduction were less frequent with dirozalkib than with crizotinib. Discontinuation due to treatment-related AEs was rare, occurring in only 1.5% of dirozalkib-treated participants, highlighting the favorable tolerability of dirozalkib during extended treatment.
The predominant TEAEs in the dirozalkib group were gastrointestinal disorders (including diarrhea, nausea, and vomiting) and elevated aminotransferase levels, most of which were grade 1 or 2 and were readily managed with supportive care or temporary dose adjustments. Diarrhea, the most frequently reported AE, typically occurred early in treatment and was transient (median duration, approximately 2 d) and mostly mild. Grade 3 diarrhea occurred in 19.0% of patients, with no grade 4 events reported, and rarely led to treatment discontinuation.
Hepatotoxicity, a common feature of ALK inhibitors [26,27], manifested as increased alanine and aspartate aminotransferase levels in 62.0% and 59.1% of dirozalkib-treated patients, respectively. Grade ≥3 transaminase elevations were infrequent in both groups (5.8% with dirozalkib versus 5.1% with crizotinib) and required fewer dose modifications with dirozalkib than with crizotinib. Hyperglycemia occurred more frequently with dirozalkib (33.6%), typically lasting approximately 30 d, whereas grade 3 to 4 treatment-related hyperglycemia was rare.
These findings are consistent with the safety profiles reported for other second-generation ALK inhibitors, which have shown comparable or higher incidences of hepatic and metabolic adverse effects [28].
Several limitations should be acknowledged. First, OS data remain immature, and the availability of effective subsequent therapies, including next-generation ALK inhibitors, may attenuate differences in OS between treatment groups, precluding definitive conclusions regarding long-term survival benefit. Thus, longer follow-up and future OS analyses will be essential to fully characterize the clinical benefit of dirozalkib.
However, the observed magnitude of the PFS advantage suggests that a future OS benefit is plausible. Notably, although the primary end point was investigator-assessed PFS, which may be subject to assessment bias, the concordant BICR-assessed PFS findings support the reliability of the PFS results. In addition, the exclusive enrollment of patients at Chinese centers limits the generalizability of the findings to other populations, underscoring the need for global validation in future studies.
The mITT analysis was conducted post hoc based on the regulatory approval status of the NGS assays used for ALK testing. This nonprespecified population may introduce selection bias. The analysis of intracranial efficacy was based on a small subset of patients with measurable brain metastases at baseline, limiting the statistical robustness and generalizability of these findings.
Finally, although the statistical analyses were not independently verified by an external party, they adhered to the prespecified statistical analysis plan and were conducted under rigorous internal quality-control procedures in accordance with regulatory requirements.
Conclusions
In summary, this phase III study demonstrated the superiority of dirozalkib over crizotinib in terms of PFS, with a manageable safety profile, in patients with advanced ALK-positive NSCLC who had not previously received ALK inhibitor therapy. These findings support dirozalkib as a promising first-line treatment option for patients with advanced ALK-positive NSCLC.
Ethical Approval
This protocol was approved by the ethics committee at the leading center of Jilin Province Cancer Hospital (202111-106-07) and each participating site, and all patients provided written informed consent before enrollment.