Targeted therapy changed the standard of care in ALK-dependent tumors. However, resistance remains a major challenge. Lorlatinib is a third-generation ALK inhibitor that inhibits most ALK mutants resistant to current ALK inhibitors. In this study, we utilize lorlatinib-resistant anaplastic large cell lymphoma (ALCL), non–small cell lung cancer (NSCLC), and neuroblastoma cell lines in vitro and in vivo to investigate the acquisition of resistance and its underlying mechanisms. ALCL cells acquired compound ALK mutations G1202R/G1269A and C1156F/L1198F in vitro at high drug concentrations. ALCL xenografts selected in vivo showed recurrent N1178H (5/10 mice) and G1269A (4/10 mice) mutations. Interestingly, intracellular localization of NPM/ALKN1178H skewed toward the cytoplasm in human cells, possibly mimicking overexpression. RNA sequencing of resistant cells showed significant alteration of PI3K/AKT and RAS/MAPK pathways. Functional validation by small-molecule inhibitors confirmed the involvement of these pathways in resistance to lorlatinib. NSCLC cells exposed in vitro to lorlatinib acquired hyperactivation of EGFR, which was blocked by erlotinib to restore sensitivity to lorlatinib. In neuroblastoma, whole-exome sequencing and proteomic profiling of lorlatinib-resistant cells revealed a truncating NF1 mutation and hyperactivation of EGFR and ErbB4. These data provide an extensive characterization of resistance mechanisms that may arise in different ALK-positive cancers following lorlatinib treatment.

Significance:

High-throughput genomic, transcriptomic, and proteomic profiling reveals various mechanisms by which multiple tumor types acquire resistance to the third-generation ALK inhibitor lorlatinib.

Activation of the anaplastic lymphoma kinase (ALK) is involved in the pathogenesis of different cancers, including anaplastic large cell lymphoma (ALCL), non–small cell lung cancer (NSCLC) and neuroblastoma (1). ALK inhibitors (ALKi) were developed for specific treatment of ALK-positive patients (1, 2). Crizotinib demonstrated superior activity compared with chemotherapy in NSCLC and showed exceptional response rates in refractory ALCL and inflammatory myofibroblastic tumor (IMT) patients (3–5). Unfortunately, the selection of drug-resistant clones has limited the long-term efficacy of crizotinib, especially in NSCLC (1, 6, 7). The knowledge of resistance mechanisms guided the quest for new drugs to overcome crizotinib failure. Several novel compounds were developed, by improving potency, selectivity, and brain penetration. Among these, lorlatinib (PF-06463922, a third-generation ALKi) showed activity against most drug-resistant mutants, including the highly refractory G1202R mutant (8–11). Indeed, ceritinib-resistant patient-derived cells carrying EML4/ALK mutations were shown to be sensitive to lorlatinib, while cells without ALK mutations were resistant (12), suggesting that resistance to this drug might arise from ALK-independent processes bypassing ALK dependency, as observed in a fraction of patients with NSCLC treated with other ALKi (6, 7, 13, 14). In such cases, drug combinations could provide effective therapeutic options (15, 16). On the other hand, compound mutations may also represent a big challenge, still poorly characterized. Indeed, a C1156Y/L1198F mutation was found in a patient relapsed on lorlatinib (17). Therefore, understanding the mechanisms leading to tumor escape is a key to the development of better therapeutic choices.

In this work, we investigated the spectrum of possible resistance mechanisms arising during lorlatinib treatment in ALK-dependent tumors. To this end, we kept ALCL, NSCLC, and neuroblastoma cells under selective pressure until drug-resistant clones evolved, in vitro and in vivo, from the original cell population.

Chemicals and cell lines

Lorlatinib and crizotinib were provided by Pfizer. Ceritinib, erlotinib, afatinib, alectinib, and trametinib were purchased from Selleck Chemicals. Karpas-299, SUP-M2, and HEK-293T cells were purchased from DSMZ, where they were routinely verified using genotypic and phenotypic testing to confirm their identity. H3122 and H2228 cell lines were provided by Dr. Claudia Voena (University of Turin, Turin, Italy). CLB-GA cells were provided by Dr. Valérie Combaret (Léon Bérard Cancer Centre, Lyon, France). Mycoplasma testing is routinely conducted on all cell lines in the laboratory.

Antibodies

The following antibodies were purchased from Cell Signaling Technology: ALK (31F12), EGFR, S6 ribosomal protein (RPS6), p44/42 MAPK (Erk1/2), STAT3, HER4/ErbB4 (111B2), AKT, phospho-ALK (Tyr1604), phospho-ALK (Tyr1278), phospho-EGFR (Tyr1068), phospho-RPS6 (Ser240/244), phospho-STAT3 (Tyr705), phospho-p44/42 MAPK (Erk1/2; Thr202/Tyr204), phospho-HER4/ErbB4 (Tyr1284; 21A9), phospho-AKT (Ser473), phospho-AKT (Thr308; D25E6). Anti-phosphotyrosine (PY20) and anti-actin antibodies were purchased from Sigma-Aldrich, and anti-tubulin was from Abcam. All antibodies were used at 1:1,000 dilution, except anti-actin (1:2,000).

In vitro selection of lorlatinib-resistant cells

Lorlatinib-resistant cell lines were established in vitro by exposing cells to gradually increasing drug concentrations, as described previously (18). To monitor cell culture growth, ALCL cell number and viability were tracked by Trypan Blue count every other day, while confluency was estimated under the microscope for adherent cells. Every time the cells resumed proliferation rates comparable with parental cells, drug concentration was increased.

In vivo studies

For selection of resistant tumors in vivo, 6-week-old female scid mice (C.B.17/IcrHanHsd-Prkdc) were purchased from Envigo Laboratories (San Pietro al Natisone, Udine, Italy) and kept under standard conditions following the guidelines of the University of Milano-Bicocca ethical committee for animal welfare. The protocol was approved by the Italian Ministry of Health and by the Institutional Committee for Animal Welfare. Lorlatinib was suspended in 0.5% carboxymethylcellulose/0.1% Tween80. Ten million Karpas-299 cells were injected subcutaneously into the left flank of the mice. Once tumors reached an average size of 200 mm3, mice were randomized to receive vehicle alone (4 mice) or lorlatinib (10 mice; starting dose 0.1 mg/kg), orally, twice a day. Tumor size was evaluated three times a week with a caliper, using the formula: tumor volume (mm3) = (d2 × D/2), where D is the longest and d is the shortest diameter. After 21 days, mice were shifted to receive 0.25 mg/kg twice a day. On day 37, lorlatinib was increased to 0.5 mg/kg twice a day. After that, each mouse was followed individually and dosage was increased every time the tumor relapsed or stabilized after partial regression (Supplementary Fig. S1). Treatment was stopped at three different doses: 0.5 mg/kg (4 mice), 1 mg/kg (3 mice), or 2 mg/kg (3 mice). To confirm drug resistance of NSCLC and neuroblastoma cells in vivo, subcutaneous tumors (parental or resistant, 106 cells) were established in nude mice and treated with 1 mg/kg twice a day (NSCLC) or 1.5 mg/kg twice a day (neuroblastoma) for two weeks.

PCR, detection of mutations, and next-generation sequencing

Total RNA was extracted using TRIzol (Invitrogen). Real-time quantitative PCR for NPM-ALK and ABCB1 was performed with Brilliant-III Ultra-Fast SYBR Green QPCR Master Mix (Agilent). Primers are reported in Supplementary Table S1. TaqMan qPCR for EGFR (Hs01076090_m1), ERBB4 (Hs00955522_m1), and ABCG2 (Hs01053790_m1) was performed using Brilliant-II QPCR Master Mix (Agilent) and probe mixes from Thermo Fisher Scientific. The beta-glucuronidase (GUS) gene was used as a reference (probe 5′-CCAGCACTCTCGTCGGTGACTGTTCA-3′). For mutation analysis, ALK kinase domain was amplified by High Fidelity Taq Polymerase (Roche) using primers shown in Supplementary Table S1. Purified PCR products were sent for Sanger sequencing to GATC Biotech or cloned by TOPO TA Cloning Kit (Invitrogen) and sequenced. Compound mutations were always confirmed by clonal Sanger sequencing. For ultra-deep sequencing, NPM-ALK kinase domain was amplified from parental and resistant cells using High Fidelity Taq Polymerase (Roche) and NPM-ALK KD primers (Supplementary Table S1). Amplicons were purified from agarose gel and sent to GalSeq for sequencing at 10,000× mean coverage. Fastq files were aligned onto the reference ALK transcript. Integrative Genomics Viewer (IGV; Broad Institute, Cambridge, MA) was used to visualize the data and annotate variants. Resistant cell–specific mutations were identified by filtering against parental cells’ data. Whole-exome sequencing (WES) and RNA-sequencing were performed as described previously (19). Briefly, genomic DNA was extracted from control and resistant cells using PureLink Genomic DNA Mini Kit (Invitrogen) and sent to GalSeq srl for sequencing at mean coverage 80×. Fastq files were aligned on reference human genome (hg38) and analyzed by CEQer2, an evolution of CEQer (20). Variants were called if present in >25% of resistant cells reads and <5% of control reads. Synonymous and noncoding substitutions were discarded. Variants with <20× coverage in either control or case samples were filtered out. For RNA-sequencing, total RNA was extracted from three independent vehicle-treated control K299 xenografts, and three lorlatinib-resistant tumors (AS4, BS1, and BD1) using TRIzol reagent, following the standard protocol. Samples were sent to Galseq srl for polyA selection, library preparation, and paired-end sequencing at approximately 50 million clusters per sample. Fastq sequences were aligned to the human genome (GRCh38/hg38) and raw counts were generated using STAR (21). Differential gene expression analysis was performed with DESeq2 tool (22). Functional enrichment for GO biological processes was performed with the Gene Set Enrichment Analysis software (23). Heatmaps were produced with GENE-E (Broad Institute, Cambridge, MA). Mutations were validated by Sanger sequencing, using the primers described in Supplementary Table S2. The NGS data discussed in this publication have been deposited in NCBI's Sequence Read Archive and are accessible through accession number PRJNA491639.

Western blotting and phospho-RTK array

Cell lysates were prepared in Laemmli buffer and run on SDS-PAGE with specific antibodies (Supplementary Methods). The Proteome Profiler Human Phospho-Kinase Array Kit (R&D Systems) was used to evaluate phosphorylation of 49 human receptor tyrosine kinases. For this assay, cell lysates from parental and resistant cells were prepared in RIPA buffer with protease and phosphatase inhibitors and 300 μg of total proteins were processed according to manufacturer's instructions. Specific antibodies are spotted in duplicate, with positive and negative control spots on membrane corners.

Proliferation, apoptosis, and colony assays

Cells (10,000/well) were incubated in the presence of the indicated compounds for 72 hours. Cell growth was assessed using the CellTiter 96 AQueous One Solution Cell Proliferation Assay System (Promega). Dose–response curves were generated using GraphPad Prism software. The IC50 value was calculated as the concentration inhibiting 50% of vehicle-treated control response (absolute IC50). For cells showing a bell-shaped dose–response curve, the IC50 was calculated relative to the maximal response or peak (relative IC50). The relative resistance (RR) index is defined as the fold shift of IC50 value as compared with control (24). Apoptosis was determined after 72 hours, using the eBioscience Annexin V Apoptosis Detection Kit FITC (Thermo Fisher Scientific) and analyzed on a FACSCalibur flow cytometer (BD Biosciences). For soft-agar colony assay, cells were suspended with drugs in RPMI:low-melting agarose as described previously (25). Colonies were counted after three weeks.

Immunofluorescence microscopy

Cells were washed with PBS and fixed with 4% p-formaldehyde in 0.12 mol/L sodium phosphate buffer, pH 7.4, and incubated for 1 hour with primary antibody (ALK, 31F12) diluted 1:100 in GDB buffer (0.02 mol/L sodium phosphate buffer, pH 7.4, 0.45 mol/L NaCl, 0.2% bovine gelatin) followed by 1-hour incubation with an Alexa 488-conjugated secondary anti IgG antibody. After washing with PBS, coverslips were stained with DAPI and mounted on glass slides with a 90% glycerol/PBS solution. HEK-293T cells, seeded on glass coverslips coated with poly-d-lysine (0.1 mg/mL), were transfected with pcDNA6.2_GFP-NPM/ALK, WT, or N1178H. After 72 hours, cells were washed with PBS, fixed, stained with DAPI, and directly mounted on glass slides with a 90% glycerol/PBS solution. For quantification, >300 cells in at least 10 acquired fields were blindly analyzed per sample. Images were acquired using a LSM 710 inverted confocal microscope (Carl Zeiss) and analyzed using a specific macro with ImageJ software to measure the fraction of nuclear ALK mean intensity over the total signal. Statistical analysis was performed using a one-way ANOVA test for K299 and ex vivo–derived cell lines or an unpaired two-tailed t test for HEK-293T–transfected cells.

ALK-dependent mechanisms driving resistance in ALCL cells in vitro and in vivo

Two human ALCL cell lines, Karpas-299 (K299) and SUP-M2 (SUPM2), were selected in vitro until they could propagate at 100 nmol/L lorlatinib. Both resistant lines (K299-LR100 and SUPM2-LR100) showed about 100-fold increased IC50 compared with parental cells (Fig. 1A and B). Analysis of ALK phosphorylation suggested that reactivation of ALK may account for resistance (Fig. 1C and D). To investigate how the resistant populations evolve upon further increase of drug concentration, K299-LR100 and SUPM2-LR100 cells were then challenged with higher lorlatinib doses, up to 1 μmol/L. The new K299-LR1000 and SUPM2-LR1000 cell lines displayed a drug-addicted phenotype (Fig. 1E and F), that is, their viability progressively decreased at lower lorlatinib doses (26). They showed persistent ALK phosphorylation at high drug concentrations (Fig. 1G and H) and increased NPM/ALK transcript level (Fig. 1I and J). Mutational analysis revealed a G1202R substitution in approximately 25% of the cells in both LR100 populations (Fig. 1K and L, top left charts). In addition, 15% of K299-LR100 developed a G1269A mutation (mutually exclusive with G1202R), while SUPM2-LR100 also carried a compound C1156F/L1198F mutation (25%). Interestingly, increasing lorlatinib concentration led to a progressive disappearance of the G1202R mutation in SUPM2-resistant cells, while the percentage of cells carrying the C1156F/L1198F mutation markedly increased, until they represented almost the entire SUPM2-LR1000 cell population (Fig. 1L). Similarly, within K299 cell line, the G1269A mutant gradually emerged as the predominant clone, upon lorlatinib increase (Fig. 1K). Noticeably, while the G1202R mutant tended to be outgrown by the G1269A clone up to 0.5 μmol/L lorlatinib, its frequency rose again at the highest dose (1 μmol/L) and the sum of G1269A- and G1202R-mutated reads exceeded 100%, suggesting that a compound mutation may have evolved. Clonal sequencing confirmed that the two mutations are on the same allele. Thus, the G1269A clone had independently acquired a second G1202R hit. No single G1202R mutant was found in K299-LR1000 cells. Overall, the identified NPM/ALK mutations may well explain resistance in these cells. These data suggest that, despite the anticipated pan-ALK–inhibitory potency of lorlatinib, in vitro–selected ALCL cell lines quickly develop ALK mutants. At high concentrations, a highly resistant double mutant clone becomes predominant. Overexpression of mutated NPM/ALK is also likely to contribute to both drug resistance and drug dependency.

Figure 1.

Lorlatinib-resistant ALCL in vitro. A–D, Characterization of LR100-resistant cells: proliferation (A and B) and Western blot analysis (C and D) of K299 (A and C) and SUPM2 (B and D) parental versus resistant cells. E–H, Characterization of LR1000-resistant cells: proliferation (E and F) and Western blot analysis (G and H) of K299 (E and G) and SUPM2 (F and H) parental versus resistant cells. I and J, NPM/ALK expression in parental and resistant K299 (I) and SUPM2 (J) cells during selection. LR100, LR200, LR500, and LR1000 indicate cells growing at 100, 200, 500, 1,000 nmol/L lorlatinib, respectively. K and L, Deep sequencing of NPM/ALK kinase domain in resistant K299 (K) and SUPM2 (L) cells during selection.

Figure 1.

Lorlatinib-resistant ALCL in vitro. A–D, Characterization of LR100-resistant cells: proliferation (A and B) and Western blot analysis (C and D) of K299 (A and C) and SUPM2 (B and D) parental versus resistant cells. E–H, Characterization of LR1000-resistant cells: proliferation (E and F) and Western blot analysis (G and H) of K299 (E and G) and SUPM2 (F and H) parental versus resistant cells. I and J, NPM/ALK expression in parental and resistant K299 (I) and SUPM2 (J) cells during selection. LR100, LR200, LR500, and LR1000 indicate cells growing at 100, 200, 500, 1,000 nmol/L lorlatinib, respectively. K and L, Deep sequencing of NPM/ALK kinase domain in resistant K299 (K) and SUPM2 (L) cells during selection.

Close modal

Next, we assessed the emergence of lorlatinib-refractory tumors in vivo. Preliminary experiments showed that 0.5 mg/kg twice a day induces full regression of parental K299 xenografts (Fig. 2A). Subsequently, mice bearing established K299 tumors were treated with lorlatinib at a suboptimal dose of 0.1 mg/kg (Fig. 2A; Supplementary Fig. S1). After 21 days, mice were shifted to 0.25 mg/kg and tumors regressed, but then relapsed. Hence, the mice underwent successive rounds of dosage increase. Each time, the tumors progressed after initial response. Three independent groups were sacrificed at different steps (0.5, 1, or 2 mg/kg lorlatinib). Finally, lorlatinib-resistant xenografts were excised and characterized. Proliferation assays confirmed resistance to lorlatinib, although at variable degrees. Sensitivity correlated with the dose at which the animals were sacrificed (Fig. 2B). The IC50 values of the established ex vivo cell lines ranged from 6 to 175-fold the IC50 of parental cells (Fig. 2C–E; Table 1). Interestingly, in few cases, the cells showed a bell-shaped proliferation curve and a slightly drug-addicted behavior (e.g., AS4 and AS6). Therefore, these cells were kept in culture in the presence of 3–10 nmol/L lorlatinib. However, in contrast to previously described drug-addicted cells (26), NPM/ALK transcript levels were not substantially upregulated (Table 1; Supplementary Fig. S2). Analysis of NPM/ALK phosphorylation suggested that resistance was partially dependent on ALK activity, as in most resistant cell lines phospho-ALK was higher, in presence of the drug, compared with parental cells (Fig. 2F). However, in several cases, relative resistance (RR) to ALK inactivation (as determined from Western blot analysis) did not correlate with cell proliferation data, suggesting that other mechanisms may be in place (Table 1; Fig. 2). Sequencing of ALK kinase domain by ultradeep and Sanger sequencing revealed the copresence of multiple NPM/ALK–mutant clones in most xenografts. Several recurrent mutations were identified, including N1178H (5/10 mice), G1269A (4 mice), I1171T (2 mice, both from the 2 mg/kg group), G1202R (3 mice, across different dose groups), and L1196M (2 mice). In particular, G1269A was a minor subclone in the 0.5 mg/kg group, and became more frequent in animals treated at higher doses, up to 100% cells in mouse AS6. Thus, G1202R and G1269A mutations recurred both in vitro and in vivo and G1269A similarly expanded under increased lorlatinib concentrations, suggesting a possible critical role in the context of resistance to lorlatinib therapy in ALCL. In some cases the observed mutations could not explain the resistant phenotype, based on their sensitivity to lorlatinib. For example, AS4 and BS1 carried a L1196M mutation, which is considered sensitive to lorlatinib (8, 10). Indeed, NPM/ALK phosphorylation was only mildly resistant to lorlatinib in these two cell lines, but the cells showed >100 RR in cell growth assays. The BD1 cell line showed contrasting results, that is, persistence of the carboxy-terminal phospho-Tyr1604 but inhibition of phospho-Tyr1278 (the first activation loop tyrosine to be phosphorylated upon kinase activation; ref. 27), suggesting that NPM/ALK was indeed inhibited by lorlatinib, but either tyrosine 1604 is phosphorylated by other kinases in these cells, or its dephosphorylation by phosphatases is impaired or slower. This discrepancy was not observed in two other lines (AS4 and BS1) or in the control. Given the heterogeneity of NPM/ALK mutations in the relapsing tumors, ectopic cell models can help elucidate the relative contribution of single mutants within a population. Analysis of transduced BaF3 cells expressing various NPM/ALK mutants confirmed that C1156F, I1171T, G1202R, and G1269A cause a significant loss of lorlatinib sensitivity (Table 1; Supplementary Fig. S3A and S3B). However, the BaF3 cell model could not explain high resistance of cells carrying a L1196M substitution, nor the observed high prevalence of N1178H mutation, which was found in 5 of 10 mice (50%), across all three dosage groups, at relevant frequency. Minor subpopulations carried N1178H in combination with G1269A (in AD5 and AD6 xenografts at 18% and 3%, respectively) but the majority of cells only harbored a single N1178H variant. BaF3 cells expressing a NPM/ALKN1178H mutant showed a low RR index, indicating that the mutation does not confer resistance per se, in this cellular model (Supplementary Fig. S3A and S3B). Because it is unlikely that half of the animals developed this mutation by chance, we sought to determine the possible mechanism of NPM/ALKN1178H action in human cells. We noticed that this mutant has an inverted cytosol to nuclear distribution ratio, compared with normal NPM/ALK: BD1 cells, carrying a homogeneous N1178H mutation (100% frequency), showed <20% nuclear localization (Fig. 3A; Supplementary Fig. S4). The same was observed in two additional K299-derived cell lines previously selected in vitro under different inhibitors and harboring 100% N1178H mutant [K1, resistant to ASP3026 (9); K300, resistant to brigatinib (28)]. In contrast, control cells and AS4 (carrying a L1196M substitution) showed the expected ratio. AS2 cells, harboring a mixture of different mutants including an N1178H subclone, showed an intermediate localization ratio. To further validate these observations, wild-type or N1178H mutant GFP-NPM/ALK were expressed in HEK-293T cells and their localization was analyzed. Cells expressing the mutant showed stronger cytoplasmic signal, while fluorescence was barely detected in the nucleus. In contrast, wild-type NPM/ALK was equally distributed in both compartments, as expected (Fig. 3B and C).

Figure 2.

Lorlatinib-resistant ALCL in vivo. A, Tumor growth of vehicle-treated mice (black) and mice treated with increasing lorlatinib doses (blue; dosing is indicated at the top) or with upfront 0.5 mg/kg lorlatinib since day 0 (orange). Dotted lines indicate the days of drug concentration switches. Between 0.5 and 2 mg/kg, each mouse followed an own timing according to individual response to the drug (Supplementary Fig. S1). B, Correlation between ex vivo IC50 values and in vivo dose reached by each xenograft. C–F,Ex vivo characterization of resistant xenografts: dose–response lorlatinib curves (C–E) grouped by dose reached in vivo (C, 0.5 mg/kg; D, 1 mg/kg; E, 2 mg/kg) and Western blot analysis of NPM/ALK dephosphorylation (F). Densitometry analysis of p-ALK/ALK signals is shown as a percentage of untreated cells.

Figure 2.

Lorlatinib-resistant ALCL in vivo. A, Tumor growth of vehicle-treated mice (black) and mice treated with increasing lorlatinib doses (blue; dosing is indicated at the top) or with upfront 0.5 mg/kg lorlatinib since day 0 (orange). Dotted lines indicate the days of drug concentration switches. Between 0.5 and 2 mg/kg, each mouse followed an own timing according to individual response to the drug (Supplementary Fig. S1). B, Correlation between ex vivo IC50 values and in vivo dose reached by each xenograft. C–F,Ex vivo characterization of resistant xenografts: dose–response lorlatinib curves (C–E) grouped by dose reached in vivo (C, 0.5 mg/kg; D, 1 mg/kg; E, 2 mg/kg) and Western blot analysis of NPM/ALK dephosphorylation (F). Densitometry analysis of p-ALK/ALK signals is shown as a percentage of untreated cells.

Close modal
Table 1.

Ex vivo characterization of resistant xenografts

ProliferationpALK-Y1604pALK-Y1278MutationsBaF3-N/A
IDDose (mg/kg)IC50 (nmol/L)RRIC50 (nmol/L)RRIC50 (nmol/L)RRqN/A (fold)VariantFreq (%)IC50 (nmol/L)RR
CTRL Vehicle 0.33 1 0.96 1 1.6 1 1.0 WT 100 1.4 1 
AS4 0.5 26a (52) 79 (158) 1.3 1.3 1.6 1 0.8 L1196M 50 5.6 4.0 
         N1178S nd - 
AD6 0.5 10 30 5.5 5.7 nd 1.0 N1178H 54 2.9 2.1 
         P1153S nd - 
         P1153H nd - 
         N1178H/G1269A nd - 
         C1156F 183 130 
         T1151A nd - 
AS2 0.5 2.3 7 4.3 4.5 nd 1.2 N1178H 74 2.9 2.1 
         G1202R 24 79 56 
AS1 0.5 11 33 43 45 nd 1.6 N1178H 90 2.9 2.1 
         G1202R 10 79 56 
BS1 30a (38) 91 (115) 5.2 5.4 3.9 2.4 0.6 L1196M 50 5.6 4.0 
AD5 17 52 54 56 nd 1.7 N1178H 64 2.9 2.1 
         N1178H/G1269A 18 nd - 
         E1241G 18 nd - 
         N1178H/C1156Y nd - 
BD1 3.2 10 164 171 4.5 2.8 1.0 N1178H 100 2.9 2.1 
BD6 23 70 53 55 nd 0.8 G1202R 37 79 56 
         G1269A 22 140 100 
         C1156Y 2.8 2.0 
         N1178H 2.9 2.1 
         I1171T 173 123 
AD3 26a (39) 79 (118) >100 >100 nd 0.5 I1171T 82 173 123 
         I1171T/E1210K 67 48 
AS6 58a (105) 176 (318) 2.5 2.6 nd 3.0 G1269A 100 140 100 
ProliferationpALK-Y1604pALK-Y1278MutationsBaF3-N/A
IDDose (mg/kg)IC50 (nmol/L)RRIC50 (nmol/L)RRIC50 (nmol/L)RRqN/A (fold)VariantFreq (%)IC50 (nmol/L)RR
CTRL Vehicle 0.33 1 0.96 1 1.6 1 1.0 WT 100 1.4 1 
AS4 0.5 26a (52) 79 (158) 1.3 1.3 1.6 1 0.8 L1196M 50 5.6 4.0 
         N1178S nd - 
AD6 0.5 10 30 5.5 5.7 nd 1.0 N1178H 54 2.9 2.1 
         P1153S nd - 
         P1153H nd - 
         N1178H/G1269A nd - 
         C1156F 183 130 
         T1151A nd - 
AS2 0.5 2.3 7 4.3 4.5 nd 1.2 N1178H 74 2.9 2.1 
         G1202R 24 79 56 
AS1 0.5 11 33 43 45 nd 1.6 N1178H 90 2.9 2.1 
         G1202R 10 79 56 
BS1 30a (38) 91 (115) 5.2 5.4 3.9 2.4 0.6 L1196M 50 5.6 4.0 
AD5 17 52 54 56 nd 1.7 N1178H 64 2.9 2.1 
         N1178H/G1269A 18 nd - 
         E1241G 18 nd - 
         N1178H/C1156Y nd - 
BD1 3.2 10 164 171 4.5 2.8 1.0 N1178H 100 2.9 2.1 
BD6 23 70 53 55 nd 0.8 G1202R 37 79 56 
         G1269A 22 140 100 
         C1156Y 2.8 2.0 
         N1178H 2.9 2.1 
         I1171T 173 123 
AD3 26a (39) 79 (118) >100 >100 nd 0.5 I1171T 82 173 123 
         I1171T/E1210K 67 48 
AS6 58a (105) 176 (318) 2.5 2.6 nd 3.0 G1269A 100 140 100 

NOTE: For each of the established cell lines, lorlatinib dose at tumor excision, activity of lorlatinib on cell proliferation and ALK phosphorylation (IC50 and corresponding RR values are reported; RR = 1 for control), ALK mutations identified and their frequency, IC50 and RR of mutants when expressed in BaF3 cells, and relative NPM/ALK expression (qN/A; control = 1) are reported. IC50 values are expressed in nmol/L. RR values are shown in bold text.

Abbreviation: N/A, NPM/ALK.

aFor cells showing a bell-shaped dose–response curve, the relative IC50 (and corresponding RR) is shown (see Materials and Methods). Absolute IC50/RR values are reported in parentheses.

Figure 3.

Analysis of resistant ALCL xenografts. A–C, Relative ALK nuclear localization signal in parental and mutant K299-derived cells (A) and in transfected HEK-293T cells (B); representative micrographs of HEK-293TGFP-NPM/ALK cells are shown in C. ***, P < 0.001. Scale bar, 20 μm. D, Number of mutations identified in three tumors by WES, filtered against control cells data. E, MA plot of differentially regulated genes in resistant versus control samples. Red dots, significantly modulated genes. F, Heatmap of hierarchical clustering of three resistant and three control tumor mRNA profiles. G and H, Gene-set enrichment analysis of resistant versus control xenografts shows enriched AKT/mTOR and RAS/MAPK signatures in lorlatinib-resistant tumors. Heatmaps of corresponding genes are shown on the right. I, Heatmap and qPCR of selected PI3K family member genes. *, P < 0.05; **, P < 0.01. H, Western blot analysis of PI3K/AKT/mTOR, RAS/MAPK, and STAT3 pathways activation in resistant versus control xenografts.

Figure 3.

Analysis of resistant ALCL xenografts. A–C, Relative ALK nuclear localization signal in parental and mutant K299-derived cells (A) and in transfected HEK-293T cells (B); representative micrographs of HEK-293TGFP-NPM/ALK cells are shown in C. ***, P < 0.001. Scale bar, 20 μm. D, Number of mutations identified in three tumors by WES, filtered against control cells data. E, MA plot of differentially regulated genes in resistant versus control samples. Red dots, significantly modulated genes. F, Heatmap of hierarchical clustering of three resistant and three control tumor mRNA profiles. G and H, Gene-set enrichment analysis of resistant versus control xenografts shows enriched AKT/mTOR and RAS/MAPK signatures in lorlatinib-resistant tumors. Heatmaps of corresponding genes are shown on the right. I, Heatmap and qPCR of selected PI3K family member genes. *, P < 0.05; **, P < 0.01. H, Western blot analysis of PI3K/AKT/mTOR, RAS/MAPK, and STAT3 pathways activation in resistant versus control xenografts.

Close modal

ALK-independent mechanisms of resistance

Activation of PI3K/AKT and RAS/MAPK pathways in ALCL xenografts.

To investigate alternative, ALK-independent mechanisms driving resistance in the xenografts that carried ALKi-sensitive mutants and/or showed a significant fraction of wild-type ALK sequence, a global next-generation sequencing (NGS) approach was taken. Two cell lines harboring the L1196M mutation (BS1 and AS4) and one carrying an N1178H mutation (BD1) were analyzed by whole-exome and mRNA sequencing, using three control xenografts as a reference. Comparative exome data showed an average of 11 (range 7–15) mutations in resistant tumors (Fig. 3D; Supplementary Table S3). Only two mutated genes (other than ALK) were shared by at least two samples: ESYT3L296M in BS1 and AS4 cells, carrying the same NPM/ALK mutation; thus, it might represent a passenger substitution in a preexisting L1196M clone. Conversely, two different RHOBTB2 mutations arose independently in BS1 (I158V) and BD1 (V126F) cells (Supplementary Fig. S5). RHOBTB2 is a tumor suppressor that is upregulated during drug-induced apoptosis and inhibits AKT (29, 30). Thus, its inactivation may contribute to drug resistance. Valine 126 is a highly conserved residue within the GTPase domain of RHOBTB2 and its change to phenylalanine is predicted to be deleterious. Further studies are ongoing to validate the role of these substitutions in resistance to ALK inhibition.

Differential gene expression analysis from RNA-seq data revealed approximately 4,000 significantly dysregulated genes in BS1, BD1, and AS4 cells compared with controls (Fig. 3E). Unsupervised hierarchical clustering showed that the three resistant samples clustered together, separated from controls (Fig. 3F). In particular, BS1 and AS4 cells were more closely related while BD1 was a little more distant, in line with their ALK mutational status. Gene-set enrichment analysis suggested two significantly enriched types of signature in the resistant samples, pointing to PI3K/AKT/mTOR (Fig. 3G) and KRAS/MAPK pathways (Fig. 3H). Analysis of gene expression data by connectivity map tool (31) indicated two AKT inhibitors among the three top negatively connected perturbagens, suggesting that AKT inhibition may lead to reversal of the biological state encoded in the query signature (i.e., resistance to lorlatinib). When closely looking at the PI3K/AKT pathway, several members of the PI3K gene family appeared deregulated, especially in AS4 cells, including PIK3CG (encoding for the PI3K-p110γ catalytic subunit), PIK3C2G (class II PI3K-C2γ), and PIK3IP1, a negative regulator of PI3K/AKT signaling (Fig. 3I; ref. 32). At protein level, all three resistant tumors showed hyperactivated MAPK and PI3K/AKT pathways compared with control, as indicated by phosphorylated (p-)ERK1/2 and by p-S6 and p-AKT, respectively (Fig. 3J). Curiously, an inverted ratio between AKT p-Ser473 and p-Thr308 signals was noted in AS4 compared with the other two xenografts, which might imply different substrate specificity (33). In contrast, STAT3 was highly activated both in control and in resistant cells. These results suggested that MAPK and PI3K pathways might provide ALK-independent survival cues bypassing ALK inhibition.

To validate these findings, AS4 cells were treated with the pan-PI3K inhibitor pictilisib (GDC-0941), alone or in combination with lorlatinib, to see whether inhibition of PI3K pathway could overturn resistance. AS4 cells were slightly more sensitive than parental K299 to the PI3K inhibitor alone (Fig. 4A). Moreover, pictilisib was able to partially reverse resistance to lorlatinib, as shown by a significant shift of the dose–response curve (approximately 5-fold IC50 reduction, P = 0.0365; Fig. 4B). When the cells were treated with a combination of lorlatinib and pictilisib during a 7-day course, we observed a significant reduction of viability, compared with single-agent treatments (Fig. 4C). Moreover, the combination was highly effective in suppressing anchorage-independent colony growth (Fig. 4D). It has been shown that prolonged G1 arrest (pG1) sensitizes lymphoma cells to PI3K inhibition through induction of PIK3IP1 expression (34). Given the striking reduction of PIK3IP1 in AS4 cells compared with parental cells, we induced AS4 cell-cycle arrest via serum deprivation and then treated cells with vehicle or pictilisib for 72 hours after readdition of serum. After 24 hours pG1, PIK3IP1 levels increased 4-fold in AS4 cells (Fig. 4E). Treatment with the PI3K inhibitor under these conditions led to near complete suppression of cell growth (Fig. 4F) and concomitant induction of cell death (Fig. 4G and H), while pictilisib treatment in nonarrested cells had only modest effects. The second most enriched pathway in our dataset was RAS/MAPK; therefore, we investigated the activity of the MEK inhibitor trametinib in lorlatinib-resistant cells. Single-agent efficacy was comparable in parental and resistant cells (Fig. 4I). However, trametinib induced a small but significant shift in sensitivity to lorlatinib when used in combination (Fig. 4J), suggesting that MAPK pathway may partially contribute to resistance to ALK inhibition in these cells. A triple combination of lorlatinib, pictilisib, and trametinib further reduced cell growth, in an additive manner (Supplementary Fig. S6). These data suggest that resistance to lorlatinib in ALCL xenografts follows complex trajectories involving both ALK-dependent and -independent pathways, coexisting in the same cell and contributing to drug resistance.

Figure 4.

Validation of PI3K/AKT and RAS/MAPK pathways as mediators of resistance. A, Proliferation of control and AS4 cells treated with pictilisib. B, Proliferation of AS4 cells treated with lorlatinib alone or combined with 1 μmol/L pictilisib. C, Time-course of single versus combined treatment in AS4 cells (50 nmol/L lorlatinib; 1 μmol/L pictilisib). D, Colony-forming assay with AS4 cells treated with vehicle, 100 nmol/L lorlatinib, 1 μmol/L pictilisib, or combination. Colonies were counted after 21 days. Representative images are shown below. E, qPCR of PIK3IP1 in AS4 cells at baseline (Control) and after 24-hour serum starvation (pG1). F, Cell growth of AS4 cells treated with vehicle (Control), vehicle after pG1, 1 μmol/L pictilisib, or 1 μmol/L pictilisib after pG1. G and H, Apoptosis by Annexin V/propidium iodide staining in AS4 cells treated as in F; representative plots (G) and quantification (H) from two experiments are shown. I, Proliferation of control and AS4 cells treated with trametinib. J, Proliferation of AS4 cells treated with lorlatinib alone or combined with 10 nmol/L trametinib. **, P < 0.01; ****, P < 0.0001.

Figure 4.

Validation of PI3K/AKT and RAS/MAPK pathways as mediators of resistance. A, Proliferation of control and AS4 cells treated with pictilisib. B, Proliferation of AS4 cells treated with lorlatinib alone or combined with 1 μmol/L pictilisib. C, Time-course of single versus combined treatment in AS4 cells (50 nmol/L lorlatinib; 1 μmol/L pictilisib). D, Colony-forming assay with AS4 cells treated with vehicle, 100 nmol/L lorlatinib, 1 μmol/L pictilisib, or combination. Colonies were counted after 21 days. Representative images are shown below. E, qPCR of PIK3IP1 in AS4 cells at baseline (Control) and after 24-hour serum starvation (pG1). F, Cell growth of AS4 cells treated with vehicle (Control), vehicle after pG1, 1 μmol/L pictilisib, or 1 μmol/L pictilisib after pG1. G and H, Apoptosis by Annexin V/propidium iodide staining in AS4 cells treated as in F; representative plots (G) and quantification (H) from two experiments are shown. I, Proliferation of control and AS4 cells treated with trametinib. J, Proliferation of AS4 cells treated with lorlatinib alone or combined with 10 nmol/L trametinib. **, P < 0.01; ****, P < 0.0001.

Close modal

EGFR activation in ALK-positive NSCLC cells.

To broaden the landscape of molecular mechanisms that may hamper lorlatinib therapy, we selected in vitro two drug-resistant ALK-positive NSCLC cell lines, employing the usual protocol of selection (18). Two NSCLC cell lines carrying different EML4/ALK variants were used: H3122 (v1) and H2228 (v3a/b). It has been suggested that variant 3 is less sensitive to ALK kinase inhibition (35). We did not observe significant differences between the two cell lines in a 72-hour proliferation assay (Fig. 5A and B). However, when put under lorlatinib selection, H2228 developed drug resistance much more slowly (Fig. 5C). For several weeks, these cells appeared as stalled, neither dying, nor growing, until they finally started to grow at a normal pace in the presence of the drug. Ultimately, it took seven months to select a population growing at 100 nmol/L lorlatinib. In contrast, H3122 underwent a rapid selection of the most resistant subclone and reached 100 nmol/L drug in half the time. In both cases, the selected cells (LR100) were highly resistant to lorlatinib (Fig. 5A and B). To confirm resistance in vivo, H3122 and H3122-LR100 xenografts were established in mice. While parental cells rapidly regressed under lorlatinib treatment, resistant cells failed to do so (Fig. 5D). However, phosphorylation of EML4/ALK was fully inhibited by lorlatinib at low nanomolar doses in both drug-resistant cell lines, while a downstream target, such as the ribosomal protein S6 (RPS6), was phosphorylated, suggesting the activation of alternative survival pathways that bypass ALK kinase inhibition (Fig. 5E and F). Indeed, both H2228- and H3122-derived resistant cells showed activation of EGFR, which was not inhibited by lorlatinib, but could be effectively blocked by erlotinib (Fig. 5G and H). Interestingly, H3122 parental cells showed rapid activation of EGFR upon lorlatinib treatment (10 nmol/L, 4 hours), suggesting an adaptive mechanism that may explain the rapid evolution of the resistant clone in this cell line. In line with these observations, treatment of H3122-LR100 with erlotinib resensitized the cells to lorlatinib (Fig. 5I). Moreover, an upfront combined ALK/EGFR block could prevent the emergence of a resistant clone from H3122 cells, while single treatments eventually allowed cells expansion (Fig. 5J and K). In contrast, H2228-LR100 were insensitive to EGFR inhibition (Fig. 5L), which indicates that they have accumulated additional genetic lesions that may account for drug resistance. The long latency for development of resistance may be responsible for this phenotype. We asked whether the MAPK pathway might be involved. To test this hypothesis, H2228-LR100 cells were treated with trametinib in combinations with lorlatinib and erlotinib. Indeed, the triple combination caused remarkable growth inhibition (Fig. 5M). Altogether, these results show that NSCLC cells can survive lorlatinib therapy by switching to ALK independency.

Figure 5.

Lorlatinib-resistant NSCLC. A and B, Proliferation of parental and resistant H3122 (A) and H2228 (B) cells treated with lorlatinib. C, Development timeline of H3122 and H2228 lorlatinib-resistant cells. Symbols correspond to drug dose increases. D, Tumor growth of parental and resistant H3122 xenografts treated with lorlatinib. E and F, Western blot analysis of parental versus resistant H3122 (E) and H2228 (F) cells treated with lorlatinib. G and H, Phospho-EGFR activation in resistant versus parental H3122 (G) and H2228 (H) cells treated with vehicle, lorlatinib, or erlotinib. I, Proliferation of H3122 and H3122-LR100 cells treated with lorlatinib, with or without 1 μmol/L erlotinib. J, Selection of drug-resistant H3122 cells under lorlatinib (10 nmol/L), erlotinib (1 μmol/L), or combination. K, Photographs of cells taken during the selection shown in J. Scale bar, 50 μm. L, Proliferation of H2228-LR100 cells treated with lorlatinib, with or without 1 μmol/L erlotinib. M, Survival of H2228-LR100 cells treated with the indicated inhibitors (lorlatinib, 100 nmol/L; erlotinib, 1 μmol/L; trametinib, 100 nmol/L). Asterisks refer to comparisons with the triple combination (*, P < 0.05; ***, P < 0.001).

Figure 5.

Lorlatinib-resistant NSCLC. A and B, Proliferation of parental and resistant H3122 (A) and H2228 (B) cells treated with lorlatinib. C, Development timeline of H3122 and H2228 lorlatinib-resistant cells. Symbols correspond to drug dose increases. D, Tumor growth of parental and resistant H3122 xenografts treated with lorlatinib. E and F, Western blot analysis of parental versus resistant H3122 (E) and H2228 (F) cells treated with lorlatinib. G and H, Phospho-EGFR activation in resistant versus parental H3122 (G) and H2228 (H) cells treated with vehicle, lorlatinib, or erlotinib. I, Proliferation of H3122 and H3122-LR100 cells treated with lorlatinib, with or without 1 μmol/L erlotinib. J, Selection of drug-resistant H3122 cells under lorlatinib (10 nmol/L), erlotinib (1 μmol/L), or combination. K, Photographs of cells taken during the selection shown in J. Scale bar, 50 μm. L, Proliferation of H2228-LR100 cells treated with lorlatinib, with or without 1 μmol/L erlotinib. M, Survival of H2228-LR100 cells treated with the indicated inhibitors (lorlatinib, 100 nmol/L; erlotinib, 1 μmol/L; trametinib, 100 nmol/L). Asterisks refer to comparisons with the triple combination (*, P < 0.05; ***, P < 0.001).

Close modal

ErbB4 activation and NF1 loss in neuroblastoma cells.

Two non-MYCN–amplified neuroblastoma cell lines expressing full-length ALK with activating mutations, SH-SY5Y (ALKF1174L) and CLB-GA (ALKR1275Q) were tested for sensitivity to lorlatinib. SH-SY5Y proved relatively resistant to treatment (IC50 ≈300 nmol/L) and did not appear to undergo any relevant population selection up to 2 μmol/L; therefore, they were not further analyzed. In contrast, CLB-GA cells were highly sensitive to lorlatinib (Fig. 6A). After several passages in the presence of the drug, a resistant cell line was selected, showing no perturbation of cell growth up to 1 μmol/L lorlatinib in vitro (CLB-GA-LR1000; Fig. 6A). These cells grew fast under lorlatinib treatment in vivo as well, while parental tumors fully regressed (Fig. 6B). However, ALK kinase was effectively inhibited by the drug, even though basal phosphorylation was a little higher in the resistant cells (ALK dephosphorylation IC50, 41 vs. 57 nmol/L; Fig. 6C). No additional mutations were found in the ALK kinase domain and neither ALK nor MYCN expression levels were upregulated (Fig. 6C; Supplementary Fig. S7A and S7B). Interestingly, global anti-phosphotyrosine blotting showed a marked increase of tyrosine kinase cascade signal in CLB-GA-LR1000 cells, further confirmed by downstream p-S6 levels (Fig. 6C). This result suggested that an alternative tyrosine kinase may have taken over as a survival pathway. Indeed, phospho-RTK array analysis showed hyperactivation of EGFR and ErbB4 kinases in lorlatinib-resistant neuroblastoma cells (Fig. 6D), probably as a result of kinase overexpression, as EGFR and ERBB4 transcript levels were approximately 3- and 300-fold higher in CLB-GA-LR1000 than in parental cells, respectively (Supplementary Fig. S7C and S7D). These results suggest that activation of ErbB family pathway may be involved in lorlatinib resistance in neuroblastoma cells, as a bypass track. Further genetic analysis by WES revealed that CLB-GA-LR1000 cells had acquired additional mutations in important genes related to cell growth and survival (Supplementary Table S4; Supplementary Fig. S5). In particular, a heterozygous truncating NF1 mutation (Fig. 6E) is predicted to cause aberrant activation of the RAS/MAPK pathway (36). Accordingly, ERK1/2 appeared to be more active in drug-resistant compared with parental cells (Fig. 6F). Treatment of CLB-GA-LR1000 cells with the pan-ErbB inhibitor afatinib efficiently shut down ErbB4 kinase activation (Fig. 6G) but, surprisingly, had only minor effects on cell growth and survival and showed poor synergism with lorlatinib. In contrast, the MEK1/2 inhibitor trametinib blunted ERK1/2 activation (Fig. 6H) and restored full sensitivity to lorlatinib (Fig. 6I), suggesting that MAPK pathway activity contributes to drug resistance in these cells. Indeed, a marked difference was evident between the observed and the predicted effects of the combination according to Bliss independence model (Δfa = −0.34). When the cells were challenged with a three-wise drug combination, cell viability was further reduced; however, the effect of adding afatinib to a lorlatinib/trametinib treatment was only additive (Δfa = −0.01), again indicating limited contribution of EGFR/ErbB4 to drug resistance.

Figure 6.

Lorlatinib-resistant neuroblastoma. A, Proliferation of parental and resistant CLB-GA cells treated with lorlatinib. B, Tumor growth of parental and resistant CLB-GA xenografts treated with lorlatinib. C, Western blot analysis of parental versus resistant CLB-GA cells treated with lorlatinib. Normalized pALK is reported. D, Hyperactivation of EGFR and ErbB4 in CLB-GA-LR1000 cells shown by phospho-RTK arrays. Phospho-EGFR and phospho-ErbB4 spots are framed and indicated by arrows. E, Sanger sequencing chromatogram showing the NF1 truncating mutation (indicated by an arrow) found in lorlatinib-resistant cells. The corresponding sequence from parental cells is shown for a comparison. F, Phosphorylation of ERK1/2 in parental and resistant CLB-GA cells; normalized quantification is shown below. G, Western blot analysis confirmation of ErbB4 aberrant activation in CLB-GA-LR1000 cells and its modulation by afatinib. H, ERK1/2 inhibition by trametinib or afatinib. I, Drug combinations overcome resistance to lorlatinib in CLB-GA-LR1000 cells. Dotted lines indicate the predicted effect of combinations according to the Bliss Independence model, typed using the same color as the corresponding combination histogram. When the observed residual viability of the combination is lower than that expected by Bliss model, the combination is supra-additive (i.e., synergistic). Addition of trametinib to lorlatinib (dark blue bar) restores full sensitivity in CLB-GA-LR1000 cells, while afatinib shows a limited effect. LOR*AF, lorlatinib + afatinib combo; LOR*TR, lorlatinib + trametinib combo; (LOR+TR)*AF, triple combo.

Figure 6.

Lorlatinib-resistant neuroblastoma. A, Proliferation of parental and resistant CLB-GA cells treated with lorlatinib. B, Tumor growth of parental and resistant CLB-GA xenografts treated with lorlatinib. C, Western blot analysis of parental versus resistant CLB-GA cells treated with lorlatinib. Normalized pALK is reported. D, Hyperactivation of EGFR and ErbB4 in CLB-GA-LR1000 cells shown by phospho-RTK arrays. Phospho-EGFR and phospho-ErbB4 spots are framed and indicated by arrows. E, Sanger sequencing chromatogram showing the NF1 truncating mutation (indicated by an arrow) found in lorlatinib-resistant cells. The corresponding sequence from parental cells is shown for a comparison. F, Phosphorylation of ERK1/2 in parental and resistant CLB-GA cells; normalized quantification is shown below. G, Western blot analysis confirmation of ErbB4 aberrant activation in CLB-GA-LR1000 cells and its modulation by afatinib. H, ERK1/2 inhibition by trametinib or afatinib. I, Drug combinations overcome resistance to lorlatinib in CLB-GA-LR1000 cells. Dotted lines indicate the predicted effect of combinations according to the Bliss Independence model, typed using the same color as the corresponding combination histogram. When the observed residual viability of the combination is lower than that expected by Bliss model, the combination is supra-additive (i.e., synergistic). Addition of trametinib to lorlatinib (dark blue bar) restores full sensitivity in CLB-GA-LR1000 cells, while afatinib shows a limited effect. LOR*AF, lorlatinib + afatinib combo; LOR*TR, lorlatinib + trametinib combo; (LOR+TR)*AF, triple combo.

Close modal

Drug resistance is a current limitation of tyrosine kinase inhibition therapy. In highly heterogeneous disease, drug pressure selects resistant clones that cause tumor relapse. Despite these drawbacks, targeted treatments have the potential to control the disease for long time. To better exploit this possibility, we need to fully understand the routes of drug resistance and devise strategies to prevent or overcome it. This study presents an array of potential mechanisms by which ALK-dependent tumors may elude targeted inhibition by the third-generation ALKi lorlatinib (Supplementary Fig. S8). In ALCL models, ALK kinase domain mutations were observed in the majority of cases, both in vitro and in vivo. In contrast to previous drugs, mostly selecting single-point mutants, lorlatinib-resistant cells progressively accumulated compound mutations. Interestingly, we identified a C1156F/L1198F mutant, nearly identical to the C1156Y/L1198F mutation found in a NSCLC patient relapsed on lorlatinib (17). In our experiments, C1156F and C1156Y responded differently as single mutants in BaF3 cells (Supplementary Fig. S3A and S3B), but the double mutation is likely to increase resistance in both cases. A G1269A variant repeatedly emerged from K299 cells both in vitro and in mice, indicating that it may represent a major challenge to lorlatinib therapy. In BaF3 cells, this mutation conferred a significant shift in sensitivity to lorlatinib. Similarly, Zou and colleagues reported an important loss of activity on EML4/ALKG1269A phosphorylation in BaF3 cells (36). In contrast, Gainor and colleagues described G1269A as sensitive to lorlatinib (7). Despite different results in different settings, this mutant was greatly enriched in a highly resistant population growing at a drug concentration not achievable clinically (37), which speaks in favor of a considerably less sensitive variant. Its association with a second hit (G1202R) is likely to further enhance drug resistance. Thus, compound mutants are predicted to become a big complication under this potent inhibitor. While this manuscript was in preparation, a similar scenario was described in EML4/ALK–positive BaF3 cells and in patients (38). Interestingly, the authors found one patient who progressed on lorlatinib carrying a G1269A mutation, and one patient with a compound G1269A/G1202R mutation, providing clinical relevance to our findings.

An N1178H mutation was frequently found in our relapsed mice. We observed that N1178H-mutant ALCL cells accumulate more NPM/ALK kinase in the cytoplasm, as compared with wild-type cells, suggesting an altered cyto-nuclear shuttling. Whether this is due to inefficient dimerization with normal NPM1, or preferential binding to cytoplasmic proteins, is unclear at present and is under investigation. We speculate that the aberrant cytoplasmic localization of N1178H mutants functionally mimics an overexpressed fusion kinase, since the cytoplasmic fraction of NPM/ALK is thought to be the oncogenic driver (26), thus leading to resistance. This behavior may also explain the curious difference in Tyr1604 versus Tyr1278 uniquely observed in N1178H-mutated BD1 cells.

Several xenografts that relapsed in vivo under lorlatinib treatment carried ALK mutations that, alone, would not fully explain tumor resistance. Fusion transcript levels were not sufficiently high to support drug resistance. As a comparison, we earlier described brigatinib-resistant cells expressing 15- to 25-fold more transcript (28). Therefore, we sought to identify other determinants of resistance. Through an unbiased NGS approach, the PI3K/AKT/mTOR signaling was found to be commonly altered in three lorlatinib-resistant xenografts. Several catalytic and regulatory subunits of the PI3K family were deregulated, including p110γ and PI3K-C2γ, ultimately pointing to aberrant activation of proliferative and survival signals. PI3K-p110γ has been shown to control T-cell survival via activation of AKT and ERK1/2 (39). It was also found upregulated in crizotinib-resistant cells and its role in driving resistance is under investigation (manuscript in preparation). PI3K-C2γ is a class II PI3K isoform that activates AKT2 (40). Abnormal activation of PI3K/AKT signaling has been linked to drug resistance in different diseases (41–43). Indeed, a pan-PI3K inhibitor restored sensitivity to lorlatinib. Further analysis indicated also a RAS/MAPK signature upregulated in resistant tumors, leading to partial efficacy of a MEK inhibitor. The MAPK and PI3K pathways have been shown to be highly interconnected by cross-talks and feedbacks, highlighting an intricate network supporting survival signals (44). These data suggest that several mechanisms may coexist in a resistant cell, each contributing part of the phenotype, including on-target mutations that do not seem, alone, to support growth (Supplementary Fig. S8). For example, the gatekeeper L1196M is considered sensitive to lorlatinib; however, mechanistically, it shows a nonnegligible loss of sensitivity (approximately 4-fold) compared with wild-type, which may have an impact when combined with other survival means.

While ALK mutations appeared to be a driving force of drug resistance in ALCL (albeit associated with additional off-target mechanisms), NSCLC and neuroblastoma cells did not develop any ALK substitution in this setting. Rather, they seemed to prefer the activation of bypass tracks, such as EGFR, ErbB4, and RAS signals (Supplementary Fig. S8). The reason why ALCL cells acquired multiple ALK mutations while NSCLC and neuroblastoma cells did not is unclear. Either ALCL cell lines harbor a larger pool of preexisting ALK-mutated cells, or solid tumors have easier access to alternative survival pathways that relieve addiction to the primary oncogene. In patients with lung cancer progressing on crizotinib, only one-third of cases are confirmed to harbor an ALK mutation or amplification, while the remaining patients show either bypass or unknown mechanisms (6). The more potent inhibitors are employed, the more frequently ALK-independent mechanisms are likely to develop. Curiously, lorlatinib-resistant H2228 cells emerged much more slowly compared with H3122. This slow kinetics is reminiscent of the evolution of drug-tolerant cells (45). Indeed, this difference reflected on the sensitivity of the two resistant cell lines to inhibition of EGFR.

The concept of complex resistance mechanisms was again recapitulated in neuroblastoma cells, in which both EGFR and ERBB4 genes were markedly upregulated and hyperactivated. In addition, the RAS/MAPK pathway was deregulated by an NF1 truncating mutation. Consequently, a combined ALK/MEK/pan-HER inhibition approach was required to completely suppress cell growth. ERBB family signaling has been associated with resistance to ALKi in ALK-positive NSCLC (46, 47), but not in neuroblastoma. On the other hand, loss of NF1 and MAPK reactivation has been linked to chemotherapy resistance and disease outcome in patients with neuroblastoma (36, 48, 49). The combined activation of these two pathways conferred high resistance to lorlatinib. Further characterization of CLB-GA-LR1000 cells revealed an increased expression of multidrug resistance transporters, ABCB1 and ABCG2, which impaired uptake of fluorescent substrates (Supplementary Fig. S9A–S9C). However, treatment with verapamil did not impact on lorlatinib sensitivity (Supplementary Fig. S9D), in line with the proposed lack of interaction of lorlatinib with efflux pumps (8). Recently, ABCB1 was shown to limit brain accumulation of lorlatinib (50). Therefore, more studies are needed to clarify its possible involvement in resistance to lorlatinib.

In conclusion, we show here that multiple resistance mechanisms may coexist in tumors, making it difficult to devise second-line therapies. This points to the importance of strategies to prevent rather than treat resistance, to allow enough time to tumor cells to develop additional ways to resist treatment. Combinatorial approaches may be effective in such cases. However, given the complexity of potential survival pathways leading to resistance, it is hard to design preventive combinations at the personalized level. In NSCLC, the high frequency of EGFR bypass mechanisms suggests that first-line ALK/EGFR combined inhibition may be beneficial, although toxicity might be an issue. Also, the MAPK pathway appears to be a common gear contributing to ALK independence across the three different tumor types. We envisage that precise definition of drug resistance mechanisms will lead to a better control of disease.

No potential conflicts of interest were disclosed.

Conception and design: S. Redaelli, C. Gambacorti-Passerini, L. Mologni

Development of methodology: C. Mastini, L. Mologni

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Redaelli, M. Ceccon, M. Zappa, G.G. Sharma, C. Mastini, M. Mauri, M. Nigoghossian, F. Farina, L. Mologni

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Redaelli, M. Zappa, G.G. Sharma, M. Mauri, M. Nigoghossian, L. Massimino, N. Cordani, R. Piazza, L. Mologni

Writing, review, and/or revision of the manuscript: S. Redaelli, M. Ceccon, L. Massimino, F. Farina, L. Mologni

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Mologni

Study supervision: C. Gambacorti-Passerini, L. Mologni

We thank Pfizer for providing crizotinib and lorlatinib. This work was supported by AIRC (IG-14249 to C. Gambacorti-Passerini; IG-17727 to R. Piazza; fellowship to M. Ceccon), by EU H2020 Marie Skłodowska-Curie action (N. 675712) and by Fondazione Umberto Veronesi (fellowship 2015 to N. Cordani). Italian Association for Cancer Research (grant no. IG-14249 to C. Gambacorti-Passerini; grant no. IG-17727 to R. Piazza; fellowship to M. Ceccon); European Union (ITN-ETN grant no. 675712 to C. Gambacorti-Passerini); Fondazione Umberto Veronesi (fellowship 2015 to N. Cordani).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Sharma
GG
,
Mota
I
,
Mologni
L
,
Patrucco
E
,
Gambacorti-Passerini
C
,
Chiarle
R
. 
Tumor resistance against ALK targeted therapy-where it comes from and where it goes
.
Cancers
2018
;
10
:
pii:E62
.
2.
Mologni
L
. 
Inhibitors of the anaplastic lymphoma kinase
.
Expert Opin Investig Drugs
2012
;
21
:
985
94
.
3.
Shaw
AT
,
Kim
DW
,
Nakagawa
K
,
Seto
T
,
Crino
L
,
Ahn
MJ
, et al
Crizotinib versus chemotherapy in advanced ALK-positive lung cancer
.
N Engl J Med
2013
;
368
:
2385
94
.
4.
Gambacorti Passerini
C
,
Farina
F
,
Stasia
A
,
Redaelli
S
,
Ceccon
M
,
Mologni
L
, et al
Crizotinib in advanced, chemoresistant anaplastic lymphoma kinase-positive lymphoma patients
.
J Natl Cancer Inst
2014
;
106
:
djt378
.
5.
Gambacorti-Passerini
C
,
Orlov
S
,
Zhang
L
,
Braiteh
F
,
Huang
H
,
Esaki
T
, et al
Long-term effects of crizotinib in ALK-positive tumors (excluding NSCLC): A phase 1b open-label study
.
Am J Hematol
2018
;
93
:
607
14
.
6.
Shaw
AT
,
Engelman
JA
. 
ALK in lung cancer: past, present, and future
.
J Clin Oncol
2013
;
31
:
1105
11
.
7.
Katayama
R
,
Shaw
AT
,
Khan
TM
,
Mino-Kenudson
M
,
Solomon
BJ
,
Halmos
B
, et al
Mechanisms of acquired crizotinib resistance in ALK-rearranged lung cancers
.
Sci Transl Med
2012
;
4
:
120ra17
.
8.
Johnson
TW
,
Richardson
PF
,
Bailey
S
,
Brooun
A
,
Burke
BJ
,
Collins
MR
, et al
Discovery of (10R)-7-amino-12-fluoro-2,10,16-trimethyl-15-oxo-10,15,16,17-tetrahydro-2H-8,4-(m etheno)pyrazolo[4,3-h][2,5,11]-benzoxadiazacyclotetradecine-3-carbonitrile (PF-06463922), a macrocyclic inhibitor of anaplastic lymphoma kinase (ALK) and c-ros oncogene 1 (ROS1) with preclinical brain exposure and broad-spectrum potency against ALK-resistant mutations
.
J Med Chem
2014
;
57
:
4720
44
.
9.
Mologni
L
,
Ceccon
M
,
Pirola
A
,
Chiriano
G
,
Piazza
R
,
Scapozza
L
, et al
NPM/ALK mutants resistant to ASP3026 display variable sensitivity to alternative ALK inhibitors but succumb to the novel compound PF-06463922
.
Oncotarget
2015
;
6
:
5720
34
.
10.
Zou
HY
,
Friboulet
L
,
Kodack
DP
,
Engstrom
LD
,
Li
Q
,
West
M
, et al
PF-06463922, an ALK/ROS1 inhibitor, overcomes resistance to first and second generation ALK inhibitors in preclinical models
.
Cancer Cell
2015
;
28
:
70
81
.
11.
Gajdosik
Z
. 
Lorlatinib. ALK/ROS1 inhibitor, Treatment of non-small cell lung cancer
.
Drugs Fut
2016
;
41
:
715
.
12.
Gainor
JF
,
Dardaei
L
,
Yoda
S
,
Friboulet
L
,
Leshchiner
I
,
Katayama
R
, et al
Molecular mechanisms of resistance to first- and second-generation ALK inhibitors in ALK-rearranged lung cancer
.
Cancer Discov
2016
;
6
:
1118
33
.
13.
Rossing
HH
,
Grauslund
M
,
Urbanska
EM
,
Melchior
LC
,
Rask
CK
,
Costa
JC
, et al
Concomitant occurrence of EGFR (epidermal growth factor receptor) and KRAS (V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog) mutations in an ALK (anaplastic lymphoma kinase)-positive lung adenocarcinoma patient with acquired resistance to crizotinib: a case report
.
BMC Res Notes
2013
;
6
:
489
.
14.
Doebele
RC
,
Pilling
AB
,
Aisner
DL
,
Kutateladze
TG
,
Le
AT
,
Weickhardt
AJ
, et al
Mechanisms of resistance to crizotinib in patients with ALK gene rearranged non-small cell lung cancer
.
Clin Cancer Res
2012
;
18
:
1472
82
.
15.
Redaelli
S
,
Ceccon
M
,
Antolini
L
,
Rigolio
R
,
Pirola
A
,
Peronaci
M
, et al
Synergistic activity of ALK and mTOR inhibitors for the treatment of NPM-ALK positive lymphoma
.
Oncotarget
2016
;
7
:
72886
97
.
16.
Crystal
AS
,
Shaw
AT
,
Sequist
LV
,
Friboulet
L
,
Niederst
MJ
,
Lockerman
EL
, et al
Patient-derived models of acquired resistance can identify effective drug combinations for cancer
.
Science
2014
;
346
:
1480
6
.
17.
Shaw
AT
,
Friboulet
L
,
Leshchiner
I
,
Gainor
JF
,
Bergqvist
S
,
Brooun
A
, et al
Resensitization to crizotinib by the lorlatinib ALK resistance mutation L1198F
.
N Engl J Med
2016
;
374
:
54
61
.
18.
Ceccon
M
,
Mologni
L
,
Bisson
W
,
Scapozza
L
,
Gambacorti-Passerini
C
. 
Crizotinib-resistant NPM-ALK mutants confer differential sensitivity to unrelated Alk inhibitors
.
Mol Cancer Res
2013
;
11
:
122
32
.
19.
Piazza
R
,
Valletta
S
,
Winkelmann
N
,
Redaelli
S
,
Spinelli
R
,
Pirola
A
, et al
Recurrent SETBP1 mutations in atypical chronic myeloid leukemia
.
Nat Genet
2013
;
45
:
18
24
.
20.
Piazza
R
,
Magistroni
V
,
Pirola
A
,
Redaelli
S
,
Spinelli
R
,
Galbiati
M
, et al
CEQer: a graphical tool for copy number and allelic imbalance detection from whole-exome sequencing data
.
PLoS One
2013
;
8
:
e74825
.
21.
Dobin
A
,
Davis
CA
,
Schlesinger
F
,
Drenkow
J
,
Zaleski
C
,
Jha
S
, et al
STAR: ultrafast universal RNA-seq aligner
.
Bioinformatics
2013
;
29
:
15
21
.
22.
Love
MI
,
Huber
W
,
Anders
S
. 
Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
.
Genome Biol
2014
;
15
:
550
.
23.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci U S A
2005
;
102
:
15545
50
.
24.
Redaelli
S
,
Mologni
L
,
Rostagno
R
,
Piazza
R
,
Magistroni
V
,
Ceccon
M
, et al
Three novel patient-derived BCR/ABL mutants show different sensitivity to second and third generation tyrosine kinase inhibitors
.
Am J Hematol
2012
;
87
:
E125
8
.
25.
Mologni
L
,
Sala
E
,
Cazzaniga
S
,
Rostagno
R
,
Kuoni
T
,
Puttini
M
, et al
Inhibition of RET tyrosine kinase by SU5416
.
J Mol Endocrinol
2006
;
37
:
199
212
.
26.
Ceccon
M
,
Boggio Merlo
ME
,
Mologni
L
,
Poggio
T
,
Varesio
LM
,
Menotti
M
, et al
Excess of NPM-ALK oncogenic signaling promotes cellular apoptosis and drug dependency
.
Oncogene
2016
;
35
:
3854
65
.
27.
Tartari
CJ
,
Gunby
RH
,
Coluccia
AM
,
Sottocornola
R
,
Cimbro
B
,
Scapozza
L
, et al
Characterization of some molecular mechanisms governing autoactivation of the catalytic domain of the anaplastic lymphoma kinase
.
J Biol Chem
2008
;
283
:
3743
50
.
28.
Ceccon
M
,
Mologni
L
,
Giudici
G
,
Piazza
R
,
Pirola
A
,
Fontana
D
, et al
Treatment efficacy and resistance mechanisms using the second-generation ALK inhibitor AP26113 in human NPM-ALK-positive anaplastic large cell lymphoma
.
Mol Cancer Res
2015
;
13
:
775
83
.
29.
Freeman
SN
,
Ma
Y
,
Cress
WD
. 
RhoBTB2 (DBC2) is a mitotic E2F1 target gene with a novel role in apoptosis
.
J Biol Chem
2008
;
283
:
2353
62
.
30.
Ji
W
,
Rivero
F
. 
Atypical Rho GTPases of the RhoBTB subfamily: roles in vesicle trafficking and tumorigenesis
.
Cells
2016
;
5
:
pii:E28
.
31.
Lamb
J
. 
The Connectivity Map: a new tool for biomedical research
.
Nat Rev Cancer
2007
;
7
:
54
60
.
32.
Zhu
Z
,
He
X
,
Johnson
C
,
Stoops
J
,
Eaker
AE
,
Stoffer
DS
, et al
PI3K is negatively regulated by PIK3IP1, a novel p110 interacting protein
.
Biochem Biophys Res Commun
2007
;
358
:
66
72
.
33.
Yung
HW
,
Charnock-Jones
DS
,
Burton
GJ
. 
Regulation of AKT phosphorylation at Ser473 and Thr308 by endoplasmic reticulum stress modulates substrate specificity in a severity dependent manner
.
PLoS One
2011
;
6
:
e17894
.
34.
Chiron
D
,
Martin
P
,
Di Liberto
M
,
Huang
X
,
Ely
S
,
Lannutti
BJ
, et al
Induction of prolonged early G1 arrest by CDK4/CDK6 inhibition reprograms lymphoma cells for durable PI3Kdelta inhibition through PIK3IP1
.
Cell Cycle
2013
;
12
:
1892
900
.
35.
Heuckmann
JM
,
Balke-Want
H
,
Malchers
F
,
Peifer
M
,
Sos
ML
,
Koker
M
, et al
Differential protein stability and ALK inhibitor sensitivity of EML4-ALK fusion variants
.
Clin Cancer Res
2012
;
18
:
4682
90
.
36.
Eleveld
TF
,
Oldridge
DA
,
Bernard
V
,
Koster
J
,
Daage
LC
,
Diskin
SJ
, et al
Relapsed neuroblastomas show frequent RAS-MAPK pathway mutations
.
Nat Genet
2015
;
47
:
864
71
.
37.
Shaw
AT
,
Felip
E
,
Bauer
TM
,
Besse
B
,
Navarro
A
,
Postel-Vinay
S
, et al
Lorlatinib in non-small-cell lung cancer with ALK or ROS1 rearrangement: an international, multicentre, open-label, single-arm first-in-man phase 1 trial
.
Lancet Oncol
2017
;
18
:
1590
9
.
38.
Yoda
S
,
Lin
JJ
,
Lawrence
MS
,
Burke
BJ
,
Friboulet
L
,
Langenbucher
A
, et al
Sequential ALK inhibitors can select for lorlatinib-resistant compound ALK mutations in ALK-positive lung cancer
.
Cancer Discov
2018
;
8
:
714
29
.
39.
Sasaki
T
,
Irie-Sasaki
J
,
Jones
RG
,
Oliveira-dos-Santos
AJ
,
Stanford
WL
,
Bolon
B
, et al
Function of PI3Kgamma in thymocyte development, T cell activation, and neutrophil migration
.
Science
2000
;
287
:
1040
6
.
40.
Braccini
L
,
Ciraolo
E
,
Campa
CC
,
Perino
A
,
Longo
DL
,
Tibolla
G
, et al
PI3K-C2gamma is a Rab5 effector selectively controlling endosomal Akt2 activation downstream of insulin signalling
.
Nat Commun
2015
;
6
:
7400
.
41.
Banno
E
,
Togashi
Y
,
de Velasco
MA
,
Mizukami
T
,
Nakamura
Y
,
Terashima
M
, et al
Clinical significance of Akt2 in advanced pancreatic cancer treated with erlotinib
.
Int J Oncol
2017
;
50
:
2049
58
.
42.
Lindblad
O
,
Cordero
E
,
Puissant
A
,
Macaulay
L
,
Ramos
A
,
Kabir
NN
, et al
Aberrant activation of the PI3K/mTOR pathway promotes resistance to sorafenib in AML
.
Oncogene
2016
;
35
:
5119
31
.
43.
Shi
H
,
Hugo
W
,
Kong
X
,
Hong
A
,
Koya
RC
,
Moriceau
G
, et al
Acquired resistance and clonal evolution in melanoma during BRAF inhibitor therapy
.
Cancer Discov
2014
;
4
:
80
93
.
44.
Carracedo
A
,
Pandolfi
PP
. 
The PTEN-PI3K pathway: of feedbacks and cross-talks
.
Oncogene
2008
;
27
:
5527
41
.
45.
Hata
AN
,
Niederst
MJ
,
Archibald
HL
,
Gomez-Caraballo
M
,
Siddiqui
FM
,
Mulvey
HE
, et al
Tumor cells can follow distinct evolutionary paths to become resistant to epidermal growth factor receptor inhibition
.
Nat Med
2016
;
22
:
262
9
.
46.
Kimura
M
,
Endo
H
,
Inoue
T
,
Nishino
K
,
Uchida
J
,
Kumagai
T
, et al
Analysis of ERBB ligand-induced resistance mechanism to crizotinib by primary culture of lung adenocarcinoma with EML4-ALK fusion gene
.
J Thorac Oncol
2015
;
10
:
527
30
.
47.
Tanizaki
J
,
Okamoto
I
,
Okabe
T
,
Sakai
K
,
Tanaka
K
,
Hayashi
H
, et al
Activation of HER family signaling as a mechanism of acquired resistance to ALK inhibitors in EML4-ALK-positive non-small cell lung cancer
.
Clin Cancer Res
2012
;
18
:
6219
26
.
48.
Holzel
M
,
Huang
S
,
Koster
J
,
Ora
I
,
Lakeman
A
,
Caron
H
, et al
NF1 is a tumor suppressor in neuroblastoma that determines retinoic acid response and disease outcome
.
Cell
2010
;
142
:
218
29
.
49.
Woodfield
SE
,
Zhang
L
,
Scorsone
KA
,
Liu
Y
,
Zage
PE
. 
Binimetinib inhibits MEK and is effective against neuroblastoma tumor cells with low NF1 expression
.
BMC Cancer
2016
;
16
:
172
.
50.
Li
W
,
Sparidans
RW
,
Wang
Y
,
Lebre
MC
,
Wagenaar
E
,
Beijnen
JH
, et al
P-glycoprotein (MDR1/ABCB1) restricts brain accumulation and Cytochrome P450–3A (CYP3A) limits oral availability of the novel ALK/ROS1 inhibitor lorlatinib
.
Int J Cancer
2018
;
143
:
2029
38
.