In 2016, Gainor and colleagues analysed 103 repeat tumour biopsies obtained at progression on first- or second-generation ALK tyrosine kinase inhibitors (TKIs) from patients with ALK-rearranged lung cancer. Each inhibitor selected a distinct spectrum of ALK kinase-domain resistance mutations, with G1202R markedly enriched after second-generation therapy.[1] In 2018, Yoda and colleagues examined 20 repeat tumour biopsies obtained after progression on lorlatinib. Compound ALK mutations were identified in seven samples (35%), and longitudinal whole-exome sequencing in three patients showed stepwise accumulation of ALK mutations during successive treatment.[2] These findings show that ALK-TKI sequencing is not biologically neutral: earlier treatment reshapes the resistance profile encountered by later drugs.

Clinical evidence supports the same conclusion. In ALEX, first-line alectinib extended median progression-free survival from 10·9 months with crizotinib to 34·8 months; central nervous system (CNS) progression was the first event in 12% and 45% of patients, respectively.[3,4] At seven years in CROWN, median progression-free survival remained unreached with first-line lorlatinib, compared with 9·1 months with crizotinib; estimated seven-year progression-free survival was 55% versus 3%. Among patients with brain metastases at baseline, the estimated probability of remaining free from intracranial progression at seven years was 83% with lorlatinib.[5] Because both trials used crizotinib as the comparator, they cannot establish whether alectinib, brigatinib, or lorlatinib is the single best option for every patient. They do, however, show that potent first-line therapy with CNS activity can keep a substantial proportion of patients from requiring another systemic treatment for many years.

First-line optimisation should therefore not be organised around preserving a formally complete generational sequence. The aim should be to identify the best initial treatment for the individual patient by balancing systemic efficacy, CNS control, long-term toxicity, comorbidity, and patient preference. Current evidence does not establish one universally optimal lifetime sequence of modern ALK TKIs; neither does it support the assumption that an earlier-generation drug preserves the full efficacy of later agents unchanged for future use.

The problem becomes more difficult after progression. Once a potent third-generation ALK TKI fails, no established higher-generation inhibitor can be substituted automatically without first defining the mechanism of resistance. Lorlatinib resistance is also markedly heterogeneous. Broadly, some tumours remain ALK dependent through on-target alterations, including compound ALK mutations,[6] whereas others escape ALK dependence through off-target processes such as epithelial–mesenchymal transition or loss of NF2 function.[7] Progression on lorlatinib is therefore a clinical event, not a single biological diagnosis.

The first task after progression on a third-generation ALK TKI is to characterise the disease state underlying that progression, rather than to make another reflexive generational switch. At least three dimensions should be considered together: the pattern of progression defines where and how extensively control has been lost; the resistance mechanism identifies the pathway on which the tumour still depends; and histology determines whether lineage identity has changed. Together, these features indicate whether ALK inhibition should be continued, modified, or discontinued.

First, the pattern of progression matters. When only a few lesions progress while the remainder of the disease remains controlled, retrospective evidence in oncogene-addicted non-small-cell lung cancer supports local ablative treatment to the progressing sites while the effective TKI is continued, thereby delaying a change in systemic therapy.[8] This strategy treats spatially restricted escape without relinquishing systemic control that remains effective elsewhere.

Second, the resistance mechanism matters. When clear on-target ALK dependence persists, further ALK-directed treatment remains biologically plausible, particularly within clinical trials of inhibitors designed to overcome compound mutations. Neladalkib (NVL-655) is a brain-penetrant, highly selective ALK inhibitor; preclinical studies showed activity against both single and compound mutations containing G1202R.[9] Phase 1/2 results from ALKOVE-1 presented at the 2026 American Society of Clinical Oncology Annual Meeting showed an objective response rate of 31% by blinded independent central review among 253 patients with advanced ALK-positive non-small-cell lung cancer previously treated with an ALK TKI. Among 47 patients with detectable G1202R mutations, the response rate was 68%.[10] By contrast, widespread progression without an actionable ALK resistance mechanism generally requires chemotherapy or a mechanism-matched combination. Platinum plus pemetrexed retains modest activity after failure of second-generation ALK TKIs, but these data do not directly define the optimal treatment after first-line lorlatinib.[11]

Third, histology cannot be inferred from molecular testing alone. Small-cell and squamous transformation have been reported after ALK-TKI therapy.[12] When clinical behaviour changes abruptly, progression is discordant between lesions, or molecular testing does not explain resistance, tissue rebiopsy can reveal a transformed phenotype that requires histology-specific treatment rather than further intensification of ALK inhibition.

A tumour state defined by progression pattern, resistance mechanism, and histology is more directly relevant to treatment than line number alone. Two patients both described as receiving second-line treatment may require entirely different approaches if one has a solitary progressing brain lesion and the other has systemic progression with small-cell transformation. Conversely, patients with different treatment histories but similar current states may be candidates for similar interventions. Line of therapy describes treatment history; it is not a biological classification.

The most defensible conclusion is not that a third-generation ALK TKI has been proved by randomised sequencing trials to be the best first-line strategy for every patient. It is that optimal first-line control should not be sacrificed merely to preserve a generational sequence. Biological and clinical evidence indicates that treatment order alters the subsequent resistance profile, whereas potent first-line therapy with CNS activity can substantially delay the need for later treatment. First things first: optimal initial ALK inhibition should not be sacrificed merely to preserve a generational sequence. Second things second: once progression occurs, the next treatment should be determined by where the disease has escaped, what mechanism is driving it, and whether its histology has changed.