Cell culture for in vitro assays

For 3D pellet culture assays, bone marrow-derived hMSCs from four different donors (Lonza Verviers) were first expanded for two passages in Lonza’s MSCGM BulletKit medium and stored in liquid nitrogen. Cells were further expanded in Dulbecco’s modified Eagle medium (DMEM), 1 g l−1 glucose, 10% fetal bovine serum (FBS), 6 mM l-glutamine, 10 mM HEPES, 50 IU ml−1 penicillin, 50 μg ml−1 streptomycin and 1 ng ml−1 human bFGF (R&D Systems). For 3D cultures, passage-6 cells were seeded at 3.5 × 105 cells per well in 96-well V-bottom plates (Costar), sedimented by centrifugation (5 min; 250g) and cultured for 4 weeks in DMEM high glucose, 0.125% bovine serum albumin (BSA; Sigma–Aldrich), ITS (6.25 μg ml−1 human insulin, 6.25 μg ml−1 human transferrin and 6.25 ng ml−1 sodium selenite; Roche), 5.3 μg ml−1 linoleic acid (Sigma–Aldrich), 50 μg ml−1l-ascorbic acid phosphate (Wako Pure Chemical), 100 ng ml−1 dexamethasone (Sigma–Aldrich), 40 µg ml−1 proline (Sigma–Aldrich), 100 IU ml−1 penicillin and 100 μg ml−1 streptomycin, supplemented as indicated with LNA043 (recombinantly expressed in Chinese hamster ovary cells; Novartis) or vehicle control. The medium was changed three times per week.

The human chondrocyte cell line C-28/I2 (licensed from M. Goldring at the Massachusetts General Hospital) was expanded in growth medium containing DMEM/F-12, 10% FCS (Millipore), 50 μg ml−1l-ascorbic acid phosphate, 100 IU ml−1 penicillin and 100 μg ml−1 streptomycin. To test LNA043, ANGPTL2, ANGPTL3 or ANGPTL4 activity, cells were seeded as a monolayer at 50,000 cells per well (96-well plate format; Costar) and treated immediately for 24 h with LNA043, human ANGPTL2, human ANGPTL3, human ANGPTL4 or vehicle control in 100 μl DMEM/F-12 medium with 1% FCS (Millipore), 50 μg ml−1l-ascorbic acid phosphate, 100 IU ml−1 penicillin and 100 μg ml−1 streptomycin.

The immortalized human bone marrow-derived mesenchymal stem cells UE7T-13 (JCRB Cell Bank) were expanded in DMEM high glucose, GlutaMAX, sodium pyruvate medium supplemented with 10% FBS (Millipore), 10 mM HEPES, non-essential amino acids, 1 ng ml−1 recombinant human bFGF, 100 IU ml−1 penicillin and 100 μg ml−1 streptomycin. To study LNA043 activity, cells were seeded two-dimensionally at 20,000 cells per well on a 96-well plate (Costar) pre-coated with bovine type I collagen (Viscofan BioEngineering) placed in chondrogenic medium consisting of DMEM high glucose, GlutaMAX, 10 mM HEPES, non-essential amino acids, 0.05% bovine serum albumin (A1595; Sigma–Aldrich), ITS (5 μg ml−1 human insulin, 5 μg ml−1 human transferrin and 5 ng ml−1 selenious acid; Corning Premix), 10 nM dexamethasone (Sigma–Aldrich), 4.7 μg ml−1 linoleic acid (Sigma–Aldrich), 100 μM l-ascorbic acid phosphate, 10 ng ml−1 recombinant human TGF-β3 (Novartis), 400 ng ml−1 human BMP2 (Novartis), 100 IU ml−1 penicillin and 100 μg ml−1 streptomycin, supplemented as indicated with LNA043, poly(I:C) (Sigma–Aldrich) or vehicle control and grown for the indicated period of time. The medium was changed after 7 d.

Human primary articular chondrocytes were isolated from the femur of three patients with OA (64–82 years old; male) undergoing TKR at the Praxisklinik Rennbahn (Muttenz, Switzerland). All participants provided written informed consent before enrollment in the study and the study was approved by the local medical ethics committee (EKNZ project ID 2020-01812). Sliced cartilage was cut into small pieces and digested at 37 °C under 5% CO2 first in 1% pronase (Roche) for 30 min and then in 0.3% type II collagenase (Worthington Biochemical Corporation) for 16 h, based on a modification of the method described by Otero et al.75. The isolated chondrocytes were then grown according to Chawla et al.48 with the following modifications: straight after isolation, the cells were resuspended and plated in culture dishes at a density of 190,000 cells per cm2 in DMEM/F-12, 10% FBS (Millipore), 100 IU ml−1 penicillin and 100 μg ml−1 streptomycin for 3 d. Following trypsinization, the cells were minimally expanded for 4 d at 10,000 cells per cm2 in DMEM high glucose, GlutaMAX, sodium pyruvate medium supplemented with 10% FBS, 0.1 mM non-essential amino acids, 10 mM HEPES, 100 IU ml−1 penicillin, 100 μg ml−1 streptomycin, 5 ng ml−1 human bFGF and 1 ng ml−1 TGF-β3 (Novartis). Thereafter, chondrocyte 3D pellets were generated after centrifugation of the cells at 350g for 5 min (200,000 cells per pellet) and cultivated in chondrogenic medium containing DMEM high glucose, GlutaMAX, sodium pyruvate medium supplemented with 0.125% BSA (A4919; Sigma–Aldrich), 0.1 mM non-essential amino acids, 10 mM HEPES, 100 IU ml−1 penicillin, 100 μg ml−1 streptomycin, 10 μg ml−1 insulin (Sigma–Aldrich), 5.5 μg ml−1 transferrin (Roche), 5.5 ng ml−1 Na selenite (Roche), 4.7 μg ml−1 linoleic acid (Roche), 0.1 mM l-ascorbic acid phosphate, 0.1 μM dexamethasone (Sigma–Aldrich) and 10 ng ml−1 human TGF-β3 (Novartis) for 2 weeks. Thereafter, cells were treated with 1 ng ml−1 human recombinant IL-1β (R&D Systems) and 1 ng ml−1 human recombinant TNF-α (R&D Systems) for 3 d. The inflammatory insult was subsequently removed and chondrogenic medium (without TGF-β3) supplemented with LNA043 or vehicle was added to the cells for a recovery phase of 7 or 11 d. The medium was changed twice per week.

For the cartilage explant assay, human tibia plateaus were collected from patients with OA undergoing TKR at the Praxisklinik Rennbahn (Muttenz, Switzerland). All participants provided written informed consent before enrollment in the study and the study was approved by the local medical ethics committee (EKNZ project ID 2020-01812). Full-thickness, 4-mm biopsy punches were taken from areas with various degrees of OA-related cartilage degeneration. Each cartilage disk was cut in half. One half was treated with LNA043 at 300 μg ml−1 while the other half was treated with vehicle. All explants were cultured in Williams Medium E (PAN Biotech) with 100 IU ml−1 penicillin and 100 mg ml−1 streptomycin (Gibco) at 37 °C and under 5% CO2 for 24 h, after which supernatants were collected for collagen type II and PRG4 ELISAs. A total of 15 cartilage explants were collected from two tibia plateaus from two patients for this study.

The angiogenesis assay was performed at Reaction Biology (Freiburg, Germany) in a modified version of the protocol described by Korff and Augustin5. Briefly, human endothelial cell spheroids were prepared as described76 by pipetting 400 HUVEC cells in a hanging drop on plastic dishes to allow overnight spheroid aggregation. Some 50 HUVEC spheroids were then seeded in 0.9 ml of a collagen gel and pipetted into individual wells of a 24-well plate to allow polymerization. LNA043, ANGPTL3 or positive controls (VEGF-A and bFGF) were added after 30 min by pipetting 100 μl of a tenfold-concentrated working solution on top of the polymerized gel. The plates were incubated at 37 °C for 24 h and fixed by adding 4% paraformaldehyde (PFA; Carl Roth). Endothelial cell sprouting was quantified by measuring the sprout length and the cumulative sprout length.

Cell viability and proliferation assay

To determine cell viability, CellTiter-Glo assay (Promega) was performed according to the manufacturer’s instructions on C-28/I2 cells, cultivated immediately after seeding, for 24 h in C-28/I2 test medium supplemented with LNA043. To determine cell proliferation, CellTiter-Glo assay was performed according to the manufacturer’s instruction on UE7T-13 cells, cultivated for 3, 7, 11 and 14 d, in UE7T-13 chondrogenic medium supplemented or not with 0.5 μg ml−1 poly(I:C) or 100 or 200 μg ml−1 LNA043. Luminescence was measured on a Mithras LB 940 reader.

All cell cultures were performed at 37 °C in a humidified incubator under 5% CO2. All cell culture additives were from Life Technologies unless otherwise indicated.

IHC and Alcian blue staining of in vitro assays

After 3D culture of hMSCs, cell pellets were stored dry at −80 °C before being embedded in optimal cutting temperature matrix, cut into 6-μm cryosections, mounted on SuperFrostPlus slides (631-0108; VWR) and stored back at −80 °C. For immunostaining, cryosections were fixed for 10 min at 4 °C in 1% PFA before a heat-induced epitope retrieval (10 mM citrate buffer; pH 6.0; 95 °C; 10 min) step, followed by permeabilization with 0.05% Triton X-100 and saturation in 10% normal donkey serum (NDS; Sigma–Aldrich) for 1 h. Sections were then incubated for 2 h with mIgG1 anti-lubricin (1:500) monoclonal antibody in 1% NDS, followed by 1 h incubation with Alexa Fluor 488 donkey anti-mIgG1 in 1% NDS. Counterstaining was done with 4′,6-diamidino-2-phenylindole (DAPI) and slides were mounted with Mowiol medium (2.1 M Mowiol (Calbiochem), 2.6 M glycerol (VWR) and 100 mM Tris (pH 8.0)). Negative controls were processed in the same way, but with the corresponding isotype antibody (Supplementary Table 1). Images were prepared with an Olympus VS120 scanner and analyzed with MATLAB (R2015a; MathWorks). Alcian blue staining of cartilage glycosaminoglycans was performed by incubating 4% PFA fixed cells with 1% Alcian Blue solution (Sigma–Aldrich) for 4 h, followed by two washing steps in 3% acetic acid and two washing steps in water.

ELISA

ELISAs were performed on cell culture supernatants stored at −80 °C. The reagents used are detailed in Supplementary Table 1. Plates were pre-coated overnight at 4 °C with a monoclonal anti-human DKK1 antibody (MSD MULTI-ARRAY High Bind 96-well plates), an anti-human IL-6 antibody (MSD MULTI-ARRAY Standard 96-well plates) or an anti-collagen type II antibody (MSD MULTI-ARRAY Standard 96-well plates), followed by blocking with 1% casein in TBS (Bio-Rad) for 1 h in a ThermoMixer at 450 r.p.m. and then washed four times with 0.5× Tris-buffered saline with 0.1% Tween 20 (TBST; Sigma–Aldrich) at room temperature. Cell supernatants were added in the appropriate wells and incubated at room temperature in a ThermoMixer for 1 h at 450 r.p.m., followed by four washing steps and the addition of biotinylated anti-human DKK1, anti-human IL-6 or anti-collagen type II secondary antibody and further incubation at room temperature in a ThermoMixer for 1 h at 450 r.p.m. After four washing steps, a streptavidin sulfo-TAG solution (0.25 μg ml−1; MSD) was added and incubated at room temperature in a ThermoMixer for 30 min at 450 r.p.m. Four additional washing steps were performed before the addition of 2× read buffer (MSD) for detection of the electrochemiluminescence counts with an MSD Sector S 600 reader. For PRG4, MSD MULTI-ARRAY Standard 96-well plates were pre-coated overnight at 4 °C with an anti-human PRG4 antibody, followed by blocking with 1% casein in TBS for 1 h in a ThermoMixer at 450 r.p.m., and then washed three times with 1× TBST at room temperature. Cell supernatants were added in these wells and incubated at room temperature in a ThermoMixer for 2 h at 450 r.p.m., followed by three washing steps and the addition of biotinylated anti-human PRG4 antibody, then further incubation at room temperature in a ThermoMixer for 2 h at 450 r.p.m. The streptavidin sulfo-TAG addition and electrochemiluminescence detection steps were similar to those described above for DKK1, IL-6 and collagen type II ELISA. The COMP and leptin ELISAs were performed according to the manufacturer’s instructions.

Quantitative reverse transcription PCR-based gene expression assessment

Cell pellets or layers were lyzed in TRIzol before extracting RNA using a Direct-zol-96 RNA kit (R2056; Zymo Research). The RNA concentration and quality were determined using DropSense 96 (PerkinElmer). Total RNA was reverse transcribed with a high-capacity complementary DNA reverse transcription kit according to the manufacturer’s protocol. Gene expression analyses were performed with a QuantStudio 5 Real-Time PCR system or an ABI PRISM 7900HT Sequence Detection System (Thermo Fisher Scientific), for hMSC and UE7T-13 cells, respectively, using TaqMan Universal PCR Master Mix and the following human TaqMan probes: ACAN (Hs00153936_m1), ALPL (Hs01029144_m1), COMP (Hs00164359_m1), IL-6 (Hs00985639_m1), LEP (Hs00174877_m1) and SOX9 (Hs00165814_m1). GUSB (Hs00939627_m1) or HPRT1 (Hs02800695_m1) were used as housekeeping genes for hMSCs and UE7T-13 cells, respectively. All reagents were from Thermo Fisher Scientific, unless otherwise indicated. The relative abundance was calculated according to the formula 2−[deltaCt (GOI − HKG)], where GOI is the gene of interest and HKG is the housekeeping gene.

Integrin knockdown in C-28/I2 cells

C-28/I2 cells were plated into a T25 flask (Costar) at a density of 4.7 × 105 cells per flask in the growth medium described above. After 24 h, cells were transfected with 30 nM control siRNA (4390843; Ambion), siRNA to integrin α5 (siRNA s7547; Ambion), siRNA to integrin αv (s7547; Ambion) or a combination of siRNA to α5 and αv using Lipofectamine RNAiMAX according to the manufacturer’s instructions (Invitrogen). After 48 h, cells were detached using Accutase (Invitrogen) and seeded onto 96-well plates at a density of 50,000 cells per well in growth medium containing 1% FCS, supplemented with either LNA043 at the indicated concentrations or vehicle control, and incubated for 24 h. Cell supernatants were analyzed for DKK1 expression by ELISA as described above. In parallel, cells were washed twice with cold PBS and lysed with RIPA buffer (Pierce) containing protease and phosphatase inhibitors (Thermo Fisher Scientific) for 15 min on ice then spun at 4 °C for 10 min at 14,000g. Protein concentrations were determined using a BCA Protein Assay Kit (Pierce). Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer was added to the samples and equal amounts of proteins were separated using 4–12% SDS-polyacrylamide gels (NuPAGE Novex 4–12% Bis-Tris gels from Invitrogen) and transferred onto nitrocellulose membranes using a Trans-Blot Turbo Transfer System (Bio-Rad). Membranes were blocked in TBST and 5% (wt/vol) milk powder or 5% BSA. Primary and secondary antibodies were incubated in TBST and 5% milk or 5% BSA. Immunoreactive bands were detected by enhanced chemiluminescence using ECL Prime reagent (Amersham) and captured using a FUSION FX (Vilber) system (Supplementary Table 1). The protein band intensity was quantitated by densitometry with Fiji software.

Co-IP experiments

A total of 200–1,000 ng samples of recombinant human α5β1 integrin (rhα5β1; R&D Systems) or recombinant human αvβ3 integrin (rhαvβ3; R&D Systems) were incubated with 200 ng LNA043 in DMEM/F-12 medium (Life Technologies) for 1 h at 37 °C. Mouse monoclonal antibody to human α5β1, human αvβ3 or mIgG were coupled to magnetic beads according to the manufacturer’s instructions (Invitrogen). Samples containing the LNA043–integrin mixture were added to the magnetic bead–antibody complex for 10 min at room temperature and processed according to the manufacturer’s instructions. Immune complexes were separated by 4–12% SDS-PAGE, transferred onto nitrocellulose membranes and subjected to immunoblotting as described above (Supplementary Table 1).

SPR analysis of LNA043 binding to integrins

Binding of LNA043 and FN1 to the integrin α5β1 and of LNA043 and VTN to the integrin αvβ3 was measured on a Biacore T200 instrument. Before the experiments, the integrin analytes (α5β1 and αvβ3) were re-buffered in running buffer (1× HBS-N, 0.005% Tween 20, 1 mM MgCl2, 1 mM CaCl2 and 2 mM MnCl2) using Amicon spin columns (Millipore). A standard amine coupling procedure was applied to immobilize the molecules on CM5 chips (Cytiva). For this, LNA043 (50 μg ml−1) and VTN (5 μg ml−1) were diluted in 10 mM sodium acetate buffer at pH 4.5 and FN1 (5 μg ml−1) was diluted in 10 mM sodium acetate buffer at pH 4.0. LNA043 was immobilized at a density of ~3,000 resonance units; FN1 and VTN were immobilized at a density of ~500 resonance units on the chip (Supplementary Tables 2 and 3). At first, the chip surface was activated with 1-ethyl-3-(dimethylaminopropyl) carbodiimide/N-hydoxysuccinimide and after that the ligands were immobilized on the corresponding flow cells. Remaining active surface groups were subsequently blocked with ethanolamine. Flow cell 1 was left blank to serve as a reference surface. Afterwards, single-cycle kinetic binding data were acquired by five subsequent injections of 1:2 dilution series of analyte (concentration range = 0.6–10 μM) over all flow cells at a flow rate of 30 μl min−1 and a temperature of 37 °C. The complexes were allowed to associate and dissociate for 60 and 120 s, respectively. Zero concentration samples (blank run) were measured to allow double referencing during data evaluation. The results were analyzed with the Biacore T200 evaluation software (version 3.0). All raw data were double referenced (blank subtracted and reference flow cell subtracted). The resulting sensorgrams were fitted using a 1:1 binding model to calculate the equilibrium dissociation constants (Kd).

Rat model of post-traumatic OA

All animal procedures were performed in compliance with Animal Welfare Act regulation 9 CFR (parts 1, 2 and 3) and US regulations as outlined in the Guide for the Care and Use of Laboratory Animals and approved by the institutional animal care and use committee. These rat studies were conducted under institutional animal care and use committee protocol 18-455, approved by the Novartis Institutes for BioMedical Research (San Diego, USA) Animal Care and Use Committee.

In all studies, 3- to 4-month-old male LEW/SsNHsd rats (Envigo, USA) were subjected to knee joint surgery to completely sever the medial collateral ligament in combination with inducing a full-thickness tear in the medial meniscus to destabilize the joint so that future weight bearing would lead to rapid degeneration of the articular cartilage. Under isoflurane anesthesia, the medial collateral ligament was severed using a scalpel blade. The scalpel blade was then slipped under the patellar ligament into the synovial space and the pointed tip was used to cut the meniscus. The skin was then closed by 9-mm wound clips (Mikron Precision stainless steel AUTOCLIP clips).

Intra-articular injections were performed under isoflurane anesthesia according to the specified experimental outline. LNA043 clinical service form (CSF) or saline was injected in a volume of 25 μl in the short-term model, and LNA043 CSF or vehicle CSF in 40 μl was injected in the therapeutic model. Injection was through the skin and the patellar ligament into the intra-articular synovial space using a 30G needle attached to a 0.3 ml syringe. At the end of the study, animals were euthanized and joints were harvested, fixed in 10% formalin, decalcified with formic acid and embedded in paraffin before sectioning. Serial step coronal sections were prepared for each knee, each containing two 5-μm serial sections encompassing the entire depth of the joint. Slides were stained with Safranin-O17 with adjacent slides that could be used for collagen II/X staining if required (Supplementary Table 1). Using a modified OARSI scoring system (Supplementary Table 4), slides were graded in a blinded manner.

Minipig cartilage injury model

This study was conducted in the facility of BioAdvice (Denmark) under animal license 2017-15-0201-01187 and in accordance with the Swiss Novartis Animal Care and Use Committee-approved study plan AGR17034, the protocol and facility standard operating procedures.

Thirty-two female minipigs aged 21–26 months at the study start (38.5–55.3 kg) were purchased from Ellegaard Göttingen Minipigs (Denmark) and housed in groups exposed to light cycles of 12 h light and 12 h dark at BioAdvice. Between the day before surgery and 2 weeks after surgery, the minipigs were housed individually.

A full-depth cartilage defect of 6 mm diameter was created surgically in the right trochlea femoris as described29. In brief, after the skin incision, the medial border of the patellar tendon was exposed and the joint was entered, taking care not to damage the cartilaginous surfaces. The distal third of the femoral trochlea was exposed and a 6-mm custom-made fenestrated drill guide (Drill Bit; 6.0 mm; length = 195/170 mm, 2-flute for quick coupling; Synthes) was used to remove the cartilage to a depth close to the subchondral bone. The surgical wound was closed in three layers and a spray bandage was applied.

Dosing started 4 weeks post-surgery. Animals were randomly divided into four groups of n = 8 and the operated knee was intra-articularly injected with 1.5 ml vehicle CSF (groups 1 and 3) or LNA043 CSF (10 mg ml−1) (groups 2 and 4) under anesthesia using a 20G needle. After withdrawal of the needle, the leg was flexed about 20 times (within 1 min) to disperse the injected substance in the joint.

At 6 months (groups 1 and 2) or 12 months (groups 3 and 4), the minipigs were euthanized, synovial fluid was collected and the knee joint, synovial membrane, meniscus and ligaments were observed macroscopically to exclude any inflammatory processes. In addition, a macroscopic evaluation of the knee articular surfaces and surrounding tissues was performed using the ICRS macroscopic cartilage damage score30 (Supplementary Table 5). This inverse scoring system ranges from 0 points (excellent cartilage repair) to 20 points (cartilage defect without any repair tissue and extension into the adjacent cartilage). Femoral condyles were explanted from the right knee of each animal, fixed in 4% neutral buffered formalin and sectioned for histological assessment. Cartilage repair was assessed in Safranin-O-stained sections using a modified O’Driscoll score31 (Supplementary Table 6).

FIH study design

The study protocol (NCT02491281) was approved by the Western Institutional Review Board, Puyallup, Washington, USA. Written informed consent was obtained from all participants and the study was performed in accordance with the Declaration of Helsinki. The study started on 16 November 2015 (first visit for first patient) and was completed on 6 March 2018 (last visit for last patient). The last patient was screened on 26 January 2018. The study population comprised male and female patients with primary OA, aged 50–75 years, who were scheduled for TKR, in good general health and on stable medication within the 3 months before. Exclusion criteria were: the use of other investigational drugs within 30 d of enrollment; a history of hypersensitivity to similar drugs; the presence of inflammatory arthropathy, active acute or chronic infection or systemic cartilage disorder; previous cartilage repair surgery; any surgical therapy or local treatment administered intra-articularly into the knee within 2 months before enrollment; a body mass index of >40; the presence of uncontrolled diabetes or hyperthyroidism; large effusion in the knee; corticosteroid use by any route except topical and nasal in the 3 months before enrollment; a history or current diagnosis of cardiovascular disease; a history of malignancy within the past 5 years; pregnant or lactating women; and women of child-bearing potential.

The primary objective of this study was to evaluate the safety and tolerability of LNA043 after one intra-articular injection into the knee. Safety laboratory evaluations included hematology, clinical chemistry and urinalysis. Secondary objectives included evaluation of: the presence and persistence of LNA043 in the joint; LNA043 pharmacokinetics in the serum and concentrations in the synovial fluid; ANGPTL3 levels in the serum and synovial fluid; and immunogenicity in the serum. Exploratory objectives included assessment of the cartilage transcriptome after exposure to LNA043 or placebo, although the analysis was largely not pre-specified.

This was a double-blind study: patients, investigator staff, persons performing the assessments and data analysts remained blinded to the allocation of study treatments. A randomization list was produced under the responsibility of Novartis using a validated system that automates the random assignment of treatment arms to randomization numbers in the specified ratio.

An independent data-monitoring committee was instituted for this study to provide unbiased assessment of potential safety issues. The data-monitoring committee for the study included individuals from the sponsor and from academia with experience in the management of patients with OA.

A total of 30 patients scheduled for TKR were randomized into seven cohorts. Two patients withdrew their consent before dosing and did not receive any study drug. The 28 patients randomized and treated (100%) completed the study. Baseline clinical values and cohort demographics are reported in Fig. 4c.

Seven patients had a protocol deviation related to key procedures not performed as per the protocol. However, none of these patients was excluded from analyses, as none of the protocol deviations was considered to have any relevant impact on safety, pharmacokinetics or biomarker analysis data.

A study protocol summary and results are available at https://www.novctrd.com/ctrdweb/trialresult/trialresults/pdf?trialResultId=17358.

This trial was sponsored by Novartis Pharmaceuticals Corporation.

FIH study statistical methods

The primary endpoint of this study was safety. No formal sample size calculation was performed and no power evaluation was provided. However, statistical considerations were made related to assessment of adverse events for this study. At a sample size of three patients on active drug per cohort, an adverse event of an underlying occurrence rate of 34% or higher would have ≥70% probability that at least one patient will report an adverse event. If the underlying occurrence rate is 42% or higher then there is ≥80% probability that at least one patient will report an adverse event.

The number and percentage of patients with adverse events were summarized by treatment group and severity. All vital signs, electrocardiogram and laboratory data were summarized using descriptive statistics. Data from patients receiving the placebo were pooled. An unblinded interim analysis was conducted after all patients of cohort 4 had TKR surgery, to select the dose and timepoint before surgery for cohort 5 and 6.

Key tissue sampling procedures in the FIH study

For every patient, articular cartilage biopsies were harvested during the TKR procedure from macroscopically damaged and undamaged areas (that is, six osteochondral samples (three damaged and three undamaged) and up to three Jamshidi needle biopsies collected from the femoral condyles, tibial plateau and patella). Also, soft tissue samples (anterior cruciate ligament, meniscus and synovial membrane) were obtained. Samples were further processed to assess local drug exposure through IHC staining and bulk RNA-seq analysis using next-generation sequencing methods. Blood samples were collected at study day 1 pre-dose in all cohorts, as well as in cohorts 1–4 and 7 at 15, 120, 240 and 480 min and 4, 8 and 36 d (end of study) post-dose, in cohort 5 at 15, 120 and 240 min and 15 and 29 d (end of study) post-dose and in cohort 6 at 15, 120 and 240 min and 4, 22 and 50 d (end of study) post-dose. Synovial fluid samples were collected at the onset of surgery only in cohorts 1–4.

LNA043 and ANGPTL3 quantification in the serum and synovial fluid

Concentrations of LNA043 in serum and synovial fluid samples were determined using validated immunocapture liquid chromatography–tandem mass spectrometry assays and anti-ANGPTL3 C-terminal specific antibody (22B16; Supplementary Table 1) with trypsin digestion. The lower limits of quantification were 10 and 20 ng ml−1 LNA043 in the serum and synovial fluid, respectively.

Validated sandwich electrochemiluminescence assays were used for quantification of full-length ANGPTL3 protein in human serum and synovial fluid samples utilizing the capture mouse monoclonal anti-ANGPTL3 N-terminal specific antibody (NEG301; Supplementary Table 1) and for detection biotinylated mouse monoclonal anti-ANGPTL3 C-terminal specific antibody (22B16) in combination with Sulfo-Tag-labeled streptavidin. The lower limits of quantification were 2.13 and 2.74 ng ml−1 ANGPTL3 in the serum and synovial fluid, respectively.

Determination of anti-LNA043 antibodies in the serum

The presence of anti-LNA043 antibodies in the serum and their potential cross-reactivity to endogenous ANGPTL3 were determined pre-dose and at the end of the study with a validated bridging electrochemiluminescence assays. In this assay, anti-LNA043 present in the analyzed serum samples formed a bridging complex with biotinylated LNA043 and Sulfo-Tag-labeled LNA043 reagents (Supplementary Table 1).

IHC staining of cartilage and H-score determination

During the TKR surgery, osteochondral samples consisting of osteochondral fragments were put in 10% neutral buffered formalin for fixation. For femoral, tibial and patella specimen decalcification, the solution was exchanged every 2 h with intermittent checks by needle for the tissue softness during the first 8 h, then samples were left in the Rapid Cal (BBC Biochemical/Fisher Scientific) solution for 68 h (over the weekend), plus another two sets of 2 h periods until the desired softness of each tissue piece was reached. The total time for decalcification was at least 76 h. Decalcified femoral, trochlear tibial and patellar bone/cartilage damaged and undamaged specimens were then bisected, and two formalin-fixed, paraffin-embedded blocks were generated. Formalin-fixed, paraffin-embedded osteochondral specimens were sectioned at 4 μm onto positively charged glass slides. The ANGPTL3 IHC staining protocol is summarized in Supplementary Table 7.

RNA isolation and RNA-seq

Total RNA was extracted from knee cartilage from TKR resection specimens. The RNA quality was evaluated with an Agilent 2100 Bioanalyzer System using Eukaryote Total RNA Pico chips (Agilent Technologies). The sequencing libraries were prepared following the Pico Input Mammalian protocol of the SMARTer Stranded Total RNA-Seq Kit v2 (Takara Bio). When possible, 250 pg RNA per sample was used as input for the sequencing library preparation. In the case of a limited amount of material, the highest possible amount of the sample was used for sequencing library preparation. Libraries were clustered in a high-output flow cell using a HiSeq PE Cluster Kit v4 on a cBot (Illumina). After cluster generation, the flow cell was loaded onto a HiSeq 2500 system for sequencing using the HiSeq SBS kit v4 (Illumina). DNA was sequenced from both ends (paired end) with a read length of 76 base pairs.

Transcriptomics data analyses

Sequencing reads were aligned to the ENSEMBL Human Genome 76 reference transcriptome (http://ftp.ensembl.org/pub/grch37/release-76/) using the Bowtie 2 aligner77 and gene counts were calculated using HTseq78. Statistical analyses were performed in R (https://www.r-project.org). Quality metrics on read duplication, transcript integrity, splice junction saturation and gene body coverage were checked using the RSeQC package79, version 2.6.2 (http://rseqc.sourceforge.net/). Counts were log2 transformed using voom methods (https://rdrr.io/bioc/limma/man/voom.html) from the limma78,80 package. Principal component analysis was performed using prcomp (https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/prcomp). Statistical differential expression analysis was performed via linear modeling using edgeR81 and limma80 and packages as available from Bioconductor (https://www.bioconductor.org) after filtering for expressed genes (genes with count per million (CPM) values above 1 in at least three samples). Gender and subject were included as co-variates in the linear model. Due to the observed correlation between the transcriptome complexity and the first principal component (PC1), library complexity was also included as a coefficient. The Benjamini–Hochberg method82 for multiple testing correction was applied to both differential expression and GSEA results. Pathway analysis and GSEA were performed using the camera function from the limma package, as described by Wu et al.83 and Ritchie et al.80, using the ranked moderated t-statistics for all contrasts as input and a signature database consisting of expert-curated OA-relevant gene sets from large initiatives such as SkeletalVis (http://phenome.manchester.ac.uk/)37 or directly from the literature. For example, the OA RAAK gene sets were curated based on Ramos et al.38. Heatmaps were generated using the ComplexHeatmap R package78.

Data have been made accessible through the NCBI Gene Expression Omnibus (accession code GSE186220).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.