Executive Summary
Lorlatinib (PF-06463922) is a small molecule kinase inhibitor targeting anaplastic lymphoma kinase (ALK) and c-ROS oncogene 1 (ROS1) as well as multiple ALK kinase domain point
mutations (L1196M, G1269A, F1174L, C1156Y, L1152R, S1206Y, I1171T and 1151Tins) identified in the tumors of patients who have developed resistance to other ALK inhibitors with Ki values of between 0.005 and 0.9 nM. Lorlatinib also inhibited phosphorylation of TYK1, FER, FPS, FAK2, ACK, FAK, TRKA, TRKB, TRKC, and PTK with IC50 values of less than 25 nM, a concentration at least 19-fold lower than the average maximum serum concentration (Cmax) of unbound
lorlatinib, or ~197 n/mL (485 nM). This calculation is based on the total Cmax of lorlatinib at the recommended dose of 100 mg, 577 ng/mL, and protein binding of ~66%. Lorlatinib inhibited phosphorylation of ALKL1196M and brain-derived neurotrophic factor-activated TRKB in cell lines. Lorlatinib also reduced proliferation of cell lines harboring ALK fusion variants with or without crizotinib resistant ALK kinase domain mutations and ROS1 oncogenic fusion
mutations. Pfizer evaluated the in vivo anti-tumor activity of lorlatinib in mice implanted with tumor cell lines expressing ROS-1 fusion variants or fusions with ALK mutations, including EML4-ALK v1L1196M and EML4-ALK v1G1202R. In these studies, administration of lorlatinib resulted in tumor growth inhibition and prolonged survival in both subcutaneous and intracranial
implantation models.
To assess the safety of lorlatinib, Pfizer conducted GLP-compliant toxicology studies of up to 13-weeks in rats and dogs. In both species, major target organs included the liver (bile duct
hyperplasia, single cell necrosis and increased sinusoid cellularity with increases in AST, ALT, ALP, total bilirubin, and cholesterol) and skin (persistent findings of discoloration, swelling, inflammation, and lesions). In rats the hepatobiliary findings were accompanied by pancreatic findings of acinar atrophy and increased serum amylase. In both species animals at the high dose levels also displayed abdominal distention. The hepatobiliary findings coupled with significant increases in cholesterol (also seen clinically), abdominal distention, and skin findings suggest cholestasis and are included in the label, under Section 13.2.
In the 13-week rat study, animals received lorlatinib twice daily by oral gavage. Females
received lower doses of lorlatinib due to higher plasma exposure values. Additional targets not described above included the kidney (tubular basophilia and hyaline cast with glomerulopathy and arterial degeneration), lymph nodes (increased cellularity and hematopoiesis), and sciatic nerve (minimal axonal degeneration). In high dose females, there were also findings of
increased Anichkov cells in the heart, possibly consistent with the increase in cholesterol. Most of these findings were reversible within the recovery period.
In the 13- week dog study, animals received lorlatinib twice daily. Moribundity occurred at the
inflammation. Observations in the remaining animals included decreased activity, decreased muscle tone, tremors, and abnormal gait (resembling the signs of peripheral neuropathy observed clinically), as well as emesis, liquid feces, warm to touch, increased incidence of skin reddening, swollen skin, dry skin, lesions, and thin fur cover. Other target organs of toxicity included the lymph nodes (minimal to mild medullary plasmacytosis, inflammation, and erythrocytosis), lungs (subacute inflammation at the mid and high doses). Decreased tidal volume also occurred in a safety pharmacology study in rats and pneumonitis occurs clinically.
High dose males also exhibited GI tract toxicity, with villous atrophy, crypt hyperplasia and subacute inflammation of the ilium, duodenum, and jejunum. Toxicity of the testes (tubular atrophy), epididymides (cellular debris), and prostate (glandular atrophy) occurred in mid and high dose dogs in the 4-week study and was reversible during a 4-week recovery period.
There was a single significant metabolite of lorlatinib found clinically at levels of approximately 21% of the parent drug, M8 (PF-06895751). This metabolite was present in both rats and dogs, though at much lower levels than in humans. Pfizer conducted in vitro pharmacology studies showing that M8 did not have pharmacologic activity at clinically relevant levels, was not mutagenic in in vitro assays, and did not inhibit hERG signaling. Given this additional data and the fact that the metabolites were present at some level in animal studies, consistent with the principles in ICHS9, no additional animal toxicology studies to qualify this metabolite are warranted.
In a central nervous system (CNS) safety pharmacology study, oral administration of lorlatinib at doses ≥ 3 mg/kg in rats resulted in clear decreases in memory recall. Lorlatinib also
demonstrated high penetrance in brain tissue, inhibition of TRKA, B, and C in biochemical assays, and inhibition of ligand-mediated TRKB phosphorylation in cells at clinically achievable concentrations. Decreases in TRKB have been associated with CNS disorders including
schizophrenia and mood disorders.24-26 Clinically, CNS effects including hallucinations, mood disorders, and seizures have occurred during treatment with lorlatinib.
There is a potential for lorlatinib to adversely affect the cardiovascular parameters. In vitro studies of lorlatinib and M8 did not predict QTc prolongation, by inhibition of the hERG channel at clinically relevant concentrations; lorlatinib itself also failed to inhibit L-type calcium channel, or Nav1.5 sodium current at clinically relevant concentrations. In an in vitro
Langendorff-perfused paced heart model, there were no lorlatinib related changes to cardiac parameters such as left ventricular pressure, coronary perfusion pressure, QRS or QT intervals. There was, however, a lorlatinib related increase in PR interval at doses ≥1 µM and above, consistent with prolonged PR intervals reported clinically. In safety pharmacology studies in rats, oral
administration of a single dose of ≥10 mg/kg lorlatinib was associated with increases in systolic, diastolic, and mean blood pressure. Conscious telemetered dogs receiving a 15 mg/kg lorlatinib dose for 12 consecutive days followed by a 5-day recovery period exhibited changes in systolic blood pressure, heart rate, PR, and QRS intervals, and in fractional shortening. Decreased systolic blood pressure as well as increased heart rate occurred on Days 12, 15 and 19. An increased PR interval was observed on Days 8 and 15, as was an increased QRS interval on Days
8 and 19. Adverse cardiovascular findings were also noted in 14-day toxicology studies in rats and dogs.
Carcinogenicity studies were not conducted with lorlatinib and are not required to support the use of a drug intended to treat patients with advanced cancer. Lorlatinib was aneugenic in an in vitro micronucleus assay using TK6 cells and positive in an in vivo bone marrow micronucleus assay in rats. Lorlatinib was negative in the in vitro bacterial reverse mutation assay.
Dedicated studies to assess fertility were not conducted or required to support the
development of a drug intended to treat patients with advanced cancer. Histopathological findings in male dogs suggest the potential for transient decreases in male fertility. There were also minimal histological findings in rats at high doses. Pfizer conducted embryo-fetal
development studies in rats and rabbits. Rats received total daily lorlatinib doses of 1, 4, 15, 30 mg/kg/day on gestation days (GDs) 6-17. In pregnant rats, there was increased
post-implantation loss at doses ≥1 mg/kg/day (approximately equal to the clinical lorlatinib exposure of 5650 ng*hr/mL at the 100 mg dose) and 100% loss at ≥4 mg/kg/day. Fetuses of dams dosed at 1 mg/kg showed external and visceral malformations including medially rotated hindlimbs, a supernumerary digit on the left forepaw, gastroschisis, and interrupted aortic arch. Rabbits received total daily doses of 1, 4, 15, 30 mg/kg/day on GDs 7-19. Abortions resulting in 100%
litter loss occurred at doses ≥15 mg/kg/day (resulting in exposures approximately 3.5 times the clinical exposure). At 4 mg/kg/day (~ 0.6 times the clinical lorlatinib exposure) there was
increased post-implantation loss, and malformations including domed head, medially
malrotated forelimbs, hyperextended forepaws, high arched palate, dilation of the lateral and third ventricles of the brain, malpositioned and misshapen kidneys, and retroesophageal subclavian artery. In an investigative study in zebrafish, Pfizer showed that there was potential for developmental toxicity, but at concentrations that were above those seen clinically. Based on data from the embryo-fetal development studies and the drug’s mechanism of action, a warning for embryo-fetal toxicity is included in the label for LORBRENA. Because the drug is genotoxic, females of reproductive potential are advised to use non-hormonal contraception (see Section Additional Safety Explorations) for at least 6 months after the last dose of
LORBRENA; similarly, males with female partners of reproductive potential are advised to use contraception for at least 3 months after the final dose. No studies were conducted or required to investigate the presence of lorlatinib in milk. Because many drugs are secreted in milk, the label includes a warning not to breastfeed during treatment with LORBRENA for 7 days after the final dose, based on half-life. There are no outstanding issues from a pharmacology/toxicology perspective that would prevent the approval of LORBRENA for the treatment of patients with ALK-positive metastatic NSCLC previously treated with one or more ALK tyrosine kinase inhibitors.
Referenced NDAs, BLAs, DMFs IND 118,296, initial IND for PF-06463922
Pharmacology Primary pharmacology
A. In Vitro Studies
Pfizer evaluated the potency of PF-06463922 against ALK and ROS1, as well as a panel of ALK and ROS1 secondary kinase domain mutations identified in tumors that developed resistance to first and second line ALK inhibitors, such as crizotinib (Study #174542). Briefly, investigators engineered baculovirus constructs to express activated kinase domains from recombinant human wild type or mutant ALK, catalytic domains from wild type or fusion mutations of ROS1, and the kinase domains from several other tyrosine kinase receptors (RTRK2, TRKB) then incubated these constructs in the presence of ATP with the test compound (11-dose 3-fold serial dilutions), DMSO (negative control), or crizotinib (reference compound) for 1 hour at room temperature. Inhibition was determined after electrophoretic separation of fluorescently labeled peptide substrates and phosphorylated product. ROS1 and non-target kinases were assayed using a microfluidic mobility shift assay.
In the biochemical assays, PF-06463922 inhibited ALK and a panel of crizotinib resistant ALK mutants, including L1196M and G1269A, the most frequently identified resistance mutations found among crizotinib-resistant tumors in patients, with a potency of up to 24 times that of crizotinib (Table 4).
Table 4: Ki of PF-06463922 for target kinases in biochemical assays
(Applicant Figure reproduced from Study 174542)
PF-06463922 was further evaluated in biochemical kinase screening assays against a panel of 206 additional recombinant kinases. Pfizer selected 11 of these kinases as potentially relevant hits, based on the level of inhibition at 1 μM, and evaluated them in biochemical or cell-based assays using inhibition of ALKL1196M as a positive control. The results of these follow-up
experiments are summarized in Table 5. Targets that were inhibited by PF-06463922 with an IC50 below the Cmax of unbound lorlatinib of approximately 197 ng/mL (~485 nM) in patients treated at the once daily oral dose of 100 mg included TYK1, FER, FPS, FAK2, ACK, FAK, TRKA, TRKB, TRKC, and PTK.
Table 5: Activity of PF-06463922 in enzyme and cell-based assays
(Applicant Figure reproduced from Study 174542)
In addition to the biochemical based assays, Pfizer investigated the activity of PF-06463922 in cellbased assays. NCI-H3122 and NCI-H2228 are two human lung adenocarcinoma cell lines harboring chromosomal inversion events resulting in expression of EML4-ALK fusion protein variant 1 (v1) and variant 3 (v3), respectively. To establish cell models expressing mutant ALK, human EML4-ALK v1L1196M and EML4-ALK v1G1269A were ectopically introduced into NCI-H3122 cells. Additionally, a panel of NIH3T3 cell lines was engineered to stably express ALK fusion variants (ALKv1, -ALKv2, -ALKv3a, -ALKv3b, and KIF5B-ALK) and crizotinib resistant EML4-ALK mutants that were identified in human tumors (L1196M, G1269A, F1174L, C1156Y, L1152R, G1202R, S1206Y and 1151Tins). Of note, the G1202R and I1171T mutations have been
identified in human tumors that developed resistance to second line ALK inhibitors such as alectinib and brigatinib. Karpas 299 human anaplastic large cell lymphoma (ALCL) cells bearing the NPM-ALK fusion and NPM-ALK fusion amplification (t(2;5) chromosomal translocation and focal amplification) were also used to test the potency of PF-06463922. The cell lines were plated with dilutions of PF-06463922, crizotinib, or controls, and incubated for 72 hours. Table _ summarizes the activity of PF-06463922 against the ALK fusion variants and ALK fusion
clinically achievable concentrations; crizotinib inhibited ALK phosphorylation at higher concentrations (Table 6).
Table 6: PF-06463922 activity against ALK fusion variants and ALK fusion mutations in cell-based assays
(Applicant Figure reproduced from Study 174542)
To examine the activity of PF-06463922 on ROS1 fusion proteins, Pfizer employed HCC78 NSCLC cells harboring chromosomal translocations that result in the expression of the constitutively active SLC34A2-ROS1(s/L) fusion proteins, NIH3T3 cells engineered to express various human oncogenic ROS1 fusion gene constructs identified in human cancers (SLC34A2-ROS1(L)
(Se12;Re32), SLC34A2-ROS1(s), CD74-ROS1(s), Fig-ROS1(L) and Fig-ROS1(s)), and Ba/F3-CD74-ROS1(s) cells, engineered to express human CD74-Ba/F3-CD74-ROS1(s) fusion. PF-06463922 inhibited the phosphorylation of ROS1 fusion variants, as summarized in Table 7.
Table 7: PF-06463922 activity against ROS1 fusion phosphorylation in cell-based assays
(Applicant Figure reproduced from Study 174542)
B. In Vivo Studies
Pfizer characterized the in vivo antitumor efficacy of PF-06463922 against the previously described ALK mutations in several xenograft mouse models. In Study #020236, NIH3T3 and Ba/F3 cells were engineered to express human EML4-ALKV1G1202 and EML4- ALKI1171T
respectively, via retroviral transduction. For both studies, cells were implanted subcutaneously into the hind flank of female athymic mice and allowed to grow to the size of 320 mm3. Mice were then treated with either PF-06463922 at doses of 0.75, 2.5, 7.5, 20, or 25 mg/kg/day for 6 days via pump infusion (Study #020236) or with either PF-0646392 at doses of 0.3, 1, and 3 mg/kg/day or crizotinib at 30 mg/kg twice daily (BID) for 12 days (Study #055600). Tumor size was measured to calculate tumor growth inhibition as evidence of anti-tumor activity. Tumor ALK phosphorylation levels were assessed by ELISA.
PF-06463922 demonstrated a dose-dependent inhibition of EML4-ALKV1G1202R phosphorylation in the subcutaneous tumors, as well as dose-dependent tumor growth inhibition (Figure 1).
Figure 1: Inhibition of EML4-ALKv1G1202R tumor growth by PF-06463922 in a NSCLC model
To compare the ability of PF-06463922 and crizotinib to inhibit ALK and ALK mutations, Pfizer produced recombinant human wild-type and mutant ALK I1171T kinase domain proteins, pre-activated them via auto-phosphorylation, and treated them with increasing concentrations (0-1 µM) of each drug. PF-06463922 inhibited both wild-type ALK and mutant ALK at lower
concentrations than crizotinib (Figure 2), with IC50values of 0.65 and 7.8 nM for wild-type ALK and ALKI1171T, respectively vs. crizotinib IC50 values of 8.3 and 30.4 nM.
Figure 2: Inhibition of ALK and ALK I1171T by lorlatinib and crizotinib
(Applicant Figure reproduced from Study 055600)
Pfizer further characterized the anti-tumor activity of PF-06463922 by examining xenograft models using luciferase expressing ALK V1 and ROS fusion NSCLC cell lines that included examining brain penetration of the drug (Study #174736). Briefly, female nu/nu or scid/beige mice (5-8 weeks old) were implanted subcutaneously in the hind flank with 5 x 106 cells in 100 µL supplemented with 50% Matrigel. Tumor size was determined by caliper measurement. A subset of animals was implanted intracranially with 3 x 105 cells, and tumor size was measured by luciferase reporter activity. PF-06463922 was administered at 5 to 10 mL/kg orally or via subcutaneous pump infusion that administered a constant infusion rate to reflect dose levels of 0.06, 0.2, 0.6, 1.5 and 3 mg/kg/day for 13 days, while crizotinib was used as a reference
compound at 80 mg/kg BID. PF-06463922 demonstrated dose-dependent anti-tumor activity in both the subcutaneous (Figure 3) and intracranial models (data not shown), suggesting that PF-06463922 is brain penetrant.
Figure 3: Inhibition of EML4-ALKL1196M (left) and ROS1 (right) tumor growth following treatment with PF-06463922 in mouse models of NSCLC
(Applicant Figure reproduced from Study 174736)
Secondary pharmacology
Pfizer screened PF-06463922 for activity against a broad panel of receptors, enzymes, transporters, and ion channels in a ligand profile screen at a concentration of 10 µM (Study
#100001517). At the 10 μM concentration PF-06463922 showed less than 50% inhibition of binding or enzymatic activity against most profiled targets with the exceptions of the following enzymes: acetylcholinesterase (73.0% inhibition), AurA/Aur2 kinase (87.0% inhibition), EGFR (73.6% inhibition) and Lck (53.4% inhibition), with the IC50 values for these activities determined to be 5.3 µM, 3.8 µM, 5 µM, and 0.7 µM, respectively.
Safety pharmacology
A. Central nervous system
Pfizer evaluated the effects of PF-06463922 on the CNS using rat hippocampal slices and in vitro screening. To examine the effects of PF-06463922 on long term potentiation (LTP), (Study
#14GR133), adult male Sprague-Dawley rats were anesthetized, and brains were removed.
Slices of the brain were perfused with artificial spinal fluid. Extracellular field excitatory
postsynaptic potential (fEPSP) recordings were made from the CA1 region of the hippocampus using a multielectrode array recording system. After obtaining baseline, a vehicle control or PF-06463922 at concentrations of 0.1 and 1 µM were applied for 20 minutes followed by theta burst stimulation (TBS) at three 10 second intervals to induce LTP. The magnitude of LTP was expressed as a percentage of fEPSP amplitude from baseline. PF-06463922 did not impact long term potentiation (LTP) in the CA1 region at the concentration of 0.1 μM, but it did reduce LTP strength when tested at 1 μM, roughly twice the concentration that would be reached at the proposed human dose of 100 mg.
Pfizer also performed two in vitro screening studies that examined potential Nav 1.5 or GABAA
receptor inhibition by PF-06463922 (Study #13GR078, 13GR081). Chinese hamster ovary (CHO) cells were stably transfected with either the human Nav1.5 gene or the human GABAα1, humanized rat GABAβ2 and human GABAγ2 genes. Cells were exposed to PF-06463922 at concentrations of 1-100 µM or a positive control (propafenone or gamma-aminobutyric acid), and inhibition was determined by whole cell patch clamp channel current readings. The IC50 for Nav1.5 current inhibition by PF-06463922 was over 100 µM. PF-06463922 also showed no inhibition of the GABAA receptor in CHO cells at concentrations up to 100 µM (Study
#13GR081).
In addition to activity against ALK, PF-06463922 was also able to inhibit TrkA and TrkB
receptors, which are involved in attention and memory processes. Therefore, Pfizer examined the effects of PF-06463922 on cognitive function (Study 12GR137). Male Wistar rats were trained under the contextual renewal paradigm. The contextual renewal model is based on the premise that conditioning occurring in one context is recalled whenever the animal is
returned to that context, but not when the animal is placed into an alternative context. The methods involve conditioning and extinction in two different contexts, and the renewal of conditioned behavior is induced by return to the conditioning context. Control groups remain in the same context during conditioning, extinction, and testing. During conditioning, rats learned to associate a cue light with food pellets. After extinction with light only presentations, animals were tested for renewal of responding to the light. Performance was assessed through the expression of the conditioned response, which included approach and activity directed at the food receptacle (nose pokes). Rats received PF-06463922 at doses of 0.3, 3, 10, and 30 mg/kg, a vehicle control, or scopolamine, a positive control known to reduce memory, at doses of 0.5 and 1 mg/kg. Animals were tested 1 hour post dose. Animals were then assessed for memory recall and cue-induced renewal responding, parameters considered to be measures of cognitive function.
Oral administration of PF-06463922 at doses ≥ 3 mg/kg resulted in clear decreases in memory recall and a trend in cue-induced renewal responding, parameters that are indicative of cognitive performance (Figure 4).
Figure 4: Effects of PF-06463922 on cognitive performance parameters
(Applicant Figure reproduced from Study 12GR137)
B. Cardiovascular system
Pfizer examined the effects of PF-06463922 on cardiovascular parameters using several in vitro and in vivo safety pharmacology studies. HEK293 cells, were stably transfected to express human ether-a-go-go (hERG) potassium channels. Patch clamp testing was performed at a temperature of 33-35°C. Onset and steady state inhibition of hERG potassium current due to PF-06463922 were measured using a pulse pattern with fixed amplitudes repeated at 5 second
intervals. PF-06463922 concentrations of 10, 30, 100 and 300 µM were applied. Terfenadine was used as a positive control. In the reported results, inhibition of hERG by PF-06463922 at 30, 100 and 300 μM was statistically significant (P < 0.05) when compared with the vehicle control values. The hERG inhibition of PF-06463922 at 10 μM was not statistically significant (P < 0.05) when compared with the vehicle control values. The IC50 for the inhibitory effect of
PF-06463922 on hERG potassium current was 203.1 μM.
PF-06463922 on hERG potassium current was 203.1 μM.