NGM282 for treatment of non-alcoholic steatohepatitis: a multicentre, randomised, double-blind, placebo-controlled, phase 2 trial

Transcription

1 NGM282 for treatment of non-alcoholic steatohepatitis: a multicentre, randomised, double-blind, placebo-controlled, phase 2 trial Stephen A Harrison, Mary E Rinella, Manal F Abdelmalek, James F Trotter, Angelo H Paredes, Hays L Arnold, Marcelo Kugelmas, Mustafa R Bashir, Mark J Jaros, Lei Ling, Stephen J Rossi, Alex M DePaoli, Rohit Loomba Summary Background Non-alcoholic steatohepatitis is a chronic liver disease characterised by the presence of hepatic steatosis, inflammation, and hepatocellular injury, for which no Food and Drug Administration (FDA)-approved treatment exists. FGF19 is a hormone that regulates bile acid synthesis and glucose homoeostasis. We aimed to assess the safety and efficacy of NGM282, an engineered FGF19 analogue, for the treatment of non-alcoholic steatohepatitis. Methods In this randomised, double-blind, placebo-controlled, phase 2 study, we recruited patients aged years with biopsy-confirmed non-alcoholic steatohepatitis as defined by the non-alcoholic steatohepatitis clinical research network histological scoring system, from hospitals and gastroenterology and liver clinics in Australia and the USA. Key eligibility criteria included a non-alcoholic fatty liver disease activity score of 4 or higher, stage 1 3 fibrosis, and at least 8% liver fat content. Patients were randomly assigned (1:1:1) via a web-based system and stratified by diabetic status to receive either 3 mg or 6 mg subcutaneous NGM282 or placebo. The primary endpoint was the absolute change from baseline to week 12 in liver fat content. Responders were patients who achieved a 5% or larger reduction in absolute liver fat content as measured by MRI-proton density fat fraction. Efficacy analysis was by intention to treat. This trial is registered with ClinicalTrials.gov, number NCT Findings Between July 14, 2015, and Aug 30, 2016, 166 patients were screened across 18 sites in Australia and the USA. 82 patients were randomly assigned to receive 3 mg NGM282, 6 mg NGM282 (n=28), or placebo. At 12 weeks, 20 (74%) patients in the 3 mg dose group and 22 (79%) in the 6 mg dose group achieved at least a 5% reduction in absolute liver fat content from baseline (relative risk 10 0 [95% CI ] vs 11 4 [ ], respectively; p< for both comparisons) versus two (7%) in the placebo group. Overall, 76 (93%) of 82 patients experienced at least one adverse event, most of which were grade 1 (55 [67%]), and only five (6%) were grade 3 or worse. The most commonly ( 10%) reported adverse events were injection site reactions (28 [34%]), diarrhoea (27 [33%]), abdominal pain (15 [18%]), and nausea (14 [17%]). These adverse events were reported more frequently in the NGM282 groups compared with the placebo group. No life-threatening events or patient deaths occurred during the study. Interpretation NGM282 produced rapid and significant reductions in liver fat content with an acceptable safety profile in patients with non-alcoholic steatohepatitis. Further study of NGM282 is warranted in this patient population. Funding NGM Biopharmaceuticals. Introduction Non-alcoholic fatty liver disease encompasses a spectrum of chronic hepatic diseases ranging from simple steatosis (fatty liver) to the more aggressive non-alcoholic steatohepatitis. Non-alcoholic steatohepatitis involves a fatty liver with inflammation and hepatocellular injury (with or without fibrosis). 1 The prevalence of non-alcoholic fatty liver disease is increasing, with estimates ranging from 20% to 40% of adults in countries adopting a western diet with the disease, 10 20% of whom progress to non-alcoholic steatohepatitis. 1 Patients with non-alcoholic steatohepatitis are at increased risk of cirrhosis and hepatocellular carcinoma, and nonalcoholic steatohepatitis is projected to be the leading indication for liver transplant in ,3 Further, most patients with non-alcoholic steatohepatitis have coexisting obesity, type 2 diabetes, insulin resistance, hypertension, or dyslipidaemia. 1 Although the mechanism underlying the development and progression from simple steatosis to non-alcoholic steatohepatitis and cirrhosis is poorly understood, insulin resistance, lipotoxicity, cytokines, oxidative stress, and other inflammatory mediators are believed to promote the development of non-alcoholic steatohepatitis and its extrahepatic complications. 4 Excess lipotoxic metabolites in the liver are believed to provide the primary insult in the pathogenesis of non-alcoholic steatohepatitis, but evidence supporting a role for bile acids in the pathogenesis of liver inflammation and fibrosis is emerging. 5 Accumulation of bile acids within hepatocytes can cause mitochondrial dysfunction, endoplasmic reticulum stress, and immune cell infiltration that can Published Online March 5, S (18) See Online/Comment S (18) Radcliffe Department of Medicine, University of Oxford, Oxford, UK (Prof S A Harrison MD); Division of Gastroenterology and Hepatology, Northwestern University, Chicago, IL, USA (Prof M E Rinella MD); Duke University, Durham, NC, USA (Prof M F Abdelmalek MD); Clinical Research and Education, Texas Digestive Disease Consultants, Dallas, TX, USA (J F Trotter MD); Division of Gastroenterology and Hepatology, Brooke Army Medical Center, San Antonio, TX, USA (A H Paredes MD); Gastroenterology Consultants of San Antonio, Live Oak, TX, USA (H L Arnold MD); Hepatology, South Denver Gastroenterology, Englewood, CO, USA (M Kugelmas MD); Department of Radiology, Duke University Medical Center, Durham, NC, USA (M R Bashir MD); Summit Analytical, Denver, CO, USA (M J Jaros PhD); NGM Biopharmaceuticals, Inc, San Francisco, CA, USA (L Ling PhD, S J Rossi PharmD, A M DePaoli MD); and Non-Alcoholic Fatty Liver Disease Research Center, Division of Gastroenterology and Epidemiology, University of California San Diego, San Francisco, CA, USA (Prof R Loomba MD) Correspondence to: Dr Stephen A Harrison, Pinnacle Clinical Research Group, Live Oak, TX 78233, USA sharrison@pinnacleresearch. com Published online March 5,

2 Research in context Evidence before this study Non-alcoholic fatty liver disease includes a spectrum of chronic hepatic diseases, with non-alcoholic steatohepatitis being the most aggressive phenotype, leading to an increased risk of developing cirrhosis and hepatocellular carcinoma. To identify clinical trials of the treatment of non-alcoholic fatty liver disease and non-alcoholic steatohepatitis, we searched PubMed for English language articles published from Jan 1, 2007, to Oct 15, 2017, with the search terms, NAFLD, NASH, fatty liver, and FGF19. We found no controlled clinical trials investigating an FGF19 analogue in the treatment of non-alcoholic steatohepatitis, and therefore the current study represents a first-in-class trial in this patient population. NGM282 is a non-tumorigenic analogue of FGF19 that retains the ability to suppress CYP7A1 without activating STAT3 signalling. Based on the established preclinical biological activity, NGM282 is currently being assessed for the treatment of non-alcoholic steatohepatitis. Added value of this study Our findings show that 3 mg and 6 mg doses of NGM282 produced rapid and sustained improvements in liver fat content over 12 weeks as measured by MRI proton density fat fraction (MRI-PDFF). Hepatic imaging with MRI-PDFF is highly sensitive to changes in liver fat content compared with histological evaluation. NGM282 also significantly decreased alanine aminotransferase (ALT), aspartate aminotransferase (AST), and non-invasive serum fibrosis biomarkers (pro-c3 and enhanced liver fibrosis score). The greatest treatment effect was recorded in patients with more severe disease (high baseline liver fat content and ALT). Decreases in liver fat content were strongly correlated with decreases in 7α-hydroxy-4-cholesten-3-one (C4), ALT, AST, and increases in low-density lipoprotein cholesterol (LDL-C). Overall, these data are the most impressive to date among pharmacological agents previously and currently being tested in non-alcoholic steatohepatitis. Both doses of NGM282 were generally well tolerated with mild adverse events. Significant decreases in serum C4 concentrations, and the ensuing increases in LDL-C were recorded with NGM282, consistent with potent target engagement and inhibition of CYP7A1. Studies are ongoing to elucidate whether NGM282-related increases in LDL-C can be mitigated with concomitant statin use. Implications of all the available evidence There is an unmet need in the treatment of non-alcoholic steatohepatitis, and no Food and Drug Administration-approved treatments currently exist. The results of our phase 2 study show the efficacy of NGM282 in rapidly decreasing liver fat content and markers of inflammation and fibrosis, highlighting its therapeutic potential in non-alcoholic steatohepatitis. Larger clinical trials of longer duration are now needed to fully assess the safety and efficacy of NGM282. ultimately lead to inflammation, cell death, and liver injury. Individuals with non-alcoholic steatohepatitis are reported to have elevated hepatic and circulating concentrations of bile acids, as well as increased concentrations of faecal and urine bile acids. 6,7 Fibroblast growth factor 19 (FGF19), an endocrine gastrointestinal hormone, regulates bile acid, carbo- hydrate, and energy homoeostasis. 8,9 The hormone controls bile acid metabolism via actions on CYP7A1, the first and rate-limiting enzyme in the classic pathway of bile acid synthesis from cholesterol. FGF19 also exerts insulin-like actions on glycogen synthesis and gluconeogenesis, thus having the potential to regulate multiple pathways involved in non-alcoholic steato hepatitis pathogenesis. Notably, circulating FGF19 concentration is reduced in patients with non-alcoholic steatohepatitis, 10,11 further suggesting that dysregulated FGF19 expression might contribute to mecha nisms governing non-alcoholic steatohepatitisrelated pro gressive liver diseases. However, the therapeutic potential of FGF19 has been hindered by its hepatocarcino genicity, as shown by findings of studies in which mice expressing an FGF19 transgene developed hepatocellular carcinoma. 12 NGM282, a non-tumorigenic variant of FGF19, is a recombinant protein being studied for treatment of non-alcoholic steatohepatitis. NGM282 (also known as M70) differs from wild-type FGF19 in the amino terminus, a key region of the protein involved in receptor interactions and signalling modulation. In NGM282, a 5-aminoacid deletion (P24 S28) coupled with the substitution of three aminoacids at crucial positions (Ala30Ser, Gly31Ser, and His33Leu) within the amino terminus enable biased FGFR4 signalling so that NGM282 retains the ability to potently repress CYP7A1 expression. 13 Importantly, unlike FGF-19, NGM282 does not activate signal transducer and activator of transcription 3 (STAT3), a signalling pathway essential for FGF19-mediated hepatocarcinogenesis. 14 Furthermore, NGM282 can block hepatocarcinogenesis associated with human FGF In animal models of non-alcoholic steatohepatitis, treatment with NGM282 resulted in a rapid and robust reduction in concentrations of alanine aminotransferase (ALT) and aspartate aminotransferase (AST), as well as a clear improvement in all histological features associated with non-alcoholic steatohepatitis, including hepatic steatosis, inflammation, ballooning degeneration, and fibrosis. 15 NGM282 was safe and well tolerated in healthy volunteers and was associated with reduction in serum concentrations of 7 α-hydroxy-4-cholesten-3-one (C4), a biomarker of hepatic CYP7A1 activity Published online March 5,

3 We aimed to assess the efficacy and safety of NGM282 versus placebo in adult patients with non-alcoholic steatohepatitis using MRI-proton density fat fraction (PDFF) to assess changes in liver fat content. Methods Study design and participants In this multicentre, international, randomised, doubleblind, placebo-controlled, phase 2 trial, patients with biopsy-confirmed non-alcoholic steatohepatitis were enrolled across 18 hospitals and gastroenterology and liver clinics in Australia and the USA. The study protocol and relevant supporting data were approved by local ethics committees before study initiation. The study was done in compliance with International Conference on Harmonisation, E6 Good Clinical Practice. Dose selection was based on phase 1 ascending dose studies showing 3 mg as the lowest dose with maximal C4 reduction (unpublished data). The 12-week treatment duration was selected based on previous trials investigating liver fat content. 17,18 Patients were eligible if they met the following inclusion criteria: aged years at the time of screening; biopsyconfirmed non-alcoholic steatohepatitis diagnosis as defined by the non-alcoholic steatohepatitis clinical research network histological scoring system, 19 with a minimum non-alcoholic fatty liver disease activity score of 4 ( 1 point in each component of steatosis, lobular inflammation, and hepatocellular ballooning); stage 1, 2, or 3 fibrosis; liver fat content 8% or higher as assessed by MRI-PDFF; and elevated ALT concentrations ( 19 IU/L in women and 30 IU/L in men). These eligibility criteria were amended on Sept 2, 2015, while blinded from previous criteria of non-alcoholic fatty liver disease activity score of 5 or higher and liver fat content of at least 10%, which were deemed too restrictive. Exclusion criteria included clinically significant acute or chronic liver disease unrelated to non-alcoholic steatohepatitis; evidence of drug-induced steatohepatitis; history or presence of compensated or decompensated cirrhosis; liver transplantation; any cardiovascular event or evidence of active cardiovascular disease within 6 months of screening; and type 1 diabetes (appendix p 3). Patients were required to remain on stable regimen of their previous type 2 diabetes and lipid-lowering treatments during the study period. All patients provided written informed consent. Randomisation and masking Patients were randomly assigned (1:1:1) via Interactive Web Response System in randomly permuted blocks (block size six) to receive 3 mg or 6 mg subcutaneous NGM282 or placebo once a day. The placebo was provided as identical pre-filled syringes in identical containers See Online for appendix 166 patients screened 84 excluded 18 did not meet histology criteria 18 did not meet MRI criteria 11 withdrew consent or were lost to follow-up 37 did not meet criteria (other) 82 randomly assigned 27 assigned to and received placebo 27 assigned to and received 3 mg NGM assigned to and received 6 mg NGM282 2 discontinued treatment 1 patient choice 1 lost to follow-up 3 discontinued treatment due to adverse event but still completed EOT MRI exam 4 discontinued treatment 3 due to adverse event 1 patient choice 2 still completed EOT MRI exam 2 did not have EOT MRI 27 had both baseline and week 12/EOT MRI 27 had both baseline and week 12/EOT MRI 26 had both baseline and week 12/EOT MRI 27 included in intention-to-treat analysis 27 included in intention-to-treat analysis 28 included in intention-to-treat analysis Figure 1: Trial profile FC=liver fat content. MRI=magnetic resonance imaging. EOT=end of treatment. Published online March 5,

4 labelled with unique code numbers, in keeping with Good Manufacturing Practice for medicinal products guidelines. A master control list of the pack identification numbers and treatment was accessible by the statistician who prepared the randomisation schedule and had no other roles in the trial. The list was also provided to the emergency un-blinding service of the contract research organisation. Patients were stratified according to diabetic status (yes or no) to ensure equitable distribution across the treatment groups. Investigators, staff, patients, funder, and medical monitor remained masked throughout the study period. Placebo 3 mg NGM282 6 mg NGM282 (n=28) Age (years) 52 8 (11 3) 52 0 (7 1) 56 4 (7 8) Sex Male 7 (26%) 11 (41%) 12 (43%) Female 20 (74%) 16 (59%) 16 (57%) Race Asian 0 1 (4%) 1 (4%) Black 2 (7%) 0 0 White 25 (93%) 25 (93%) 24 (86%) Pacific Islander (4%) Other 0 1 (4%) 2 (7%) Ethnic origin Hispanic or Latino 12 (44%) 8 (30%) 8 (29%) Liver enzymes Alanine aminotransferase (IU/L) 71 1 (42) 67 4 (54) 61 8 (34) Aspartate aminotransferase (IU/L) 59 2 (30) 50 4 (32) 43 1 (22) Alkaline phosphatase (U/L) 97 7 (39 7) (67 3) 91 5 (30 5) γ-glutamyl transpeptidase (U/L) 79 5 (66 8) 97 4 (98 5) 87 5 (88 4) Total bilirubin (mg/dl) 0 3 (0 2) 0 5 (0 3) 0 4 (0 2) Lipids Cholesterol (mmol/l) 4 8 (0 9) 4 8 (1 5) 4 8 (1 2) HDL cholesterol (mmol/l) 1 0 (0 3) 1 1 (0 6) 1 0 (0 3) LDL cholesterol (mmol/l) 2 6 (0 8) 2 2 (1 1) 2 6 (1 0) Triglycerides (mg/dl) 1 9 (1 2) 2 2 (1 5) 1 8 (0 9) C4 (ng/ml) 32 0 (23 6) 49 8 (54 0) 29 4 (22 3) Hematology Hemoglobin (g/l) 13 6 (1 4) 14 7 (1 6) 14 1 (1 4) Hematocrit 42 0% (4 0) 44 7% (4 5) 43 7% (4 5) Mean corpuscular volume (FL) 91 0 (5 4) 91 0 (3 8) 89 6 (4 0) White blood cell count (K/UL) 7 3 (2 0) 7 8 (2 8) 7 4 (2 3) Platelet count (K/UL) (75 8) (59 0) (68 5) Chemistries Bicarbonate (mmol/l) 22 9 (2 6) 24 4 (2 4) 24 2 (2 3) Calcium (mg/dl ) 9 1 (0 5) 9 1 (0 5) 9 2 (0 3) Phosphate (mg/dl ) 3 4 (0 6) 3 4 (0 5) 3 4 (0 5) Creatinine (mg/dl ) 0 8 (0 2) 0 8 (0 2) 0 8 (0 2) Uric acid (mg/dl ) 5 4 (1 5) 6 7 (1 3) 5 9 (1 3) Albumin (g/dl) 4 1 (0 38) 4 3 (0 33) 4 3 (0 33) Other laboratory results INR 1 1 (0 1) 1 1 (0 1) 1 1 (0 1) (Table 1 continues on next page) Procedures During the screening period and before randomisation, patients underwent a liver biopsy or provided a liver biopsy tissue specimen obtained within the previous 6 months. Tissue samples were prepared and read by a qualified local pathologist to verify non-alcoholic steatohepatitis and determine the non-alcoholic fatty liver disease activity score (NAS) grade. On day 1, study drug self-administration instructions and training were provided to patients, and a weekly study drug kit was dispensed. The first dose of study drug and doses at weeks 1, 2, 4, 6, 8, and 12 were self-administered in the clinic; all other doses were administered at home. Patients were instructed to inject the study drug at the same time every morning. Patients underwent MRI-PDFF during screening and at the end of treatment or early withdrawal visits to assess liver fat content. Scans were performed on qualified and standardised instruments at high field strength (3T at eight sites, 1 5T at ten sites), without oral or intravenous contrast. Local MRI facilities completed a qualification and quality assurance process before performing study MRI examinations. MRI-PDFF acquisition protocols included a six-echo 2D gradient recalled echo sequence, and image data were transferred to the Center for Advanced Magnetic Resonance Development at Duke University, Durham, NC, USA, for central calculation and measurement of MRI-PDFF using an established technique. 20,21 After the end of the trial, MRI data were randomly assigned with respect to timepoint, and the central reader measured MRI-PDFF data masked to treatment group, clinical data, local site of origin, and timepoint. Patients were assessed at weeks 1, 2, 4, and 8 for on-treatment assessments, at the end of treatment on week 12, and at follow-up 4 weeks after the last dose. Bodyweight, BMI, and waist circumferences were measured at screening, day 1, and weeks 12 and end of treatment and 16. A fasting lipid panel was obtained on day 1 and weeks 4 and 12. Additionally, C4 concentrations were obtained on day 1 and week 12, and we measured haemoglobin A 1c (HbA 1c ) concentrations, insulin concentrations, and homoeostasis model assessment-estimated insulin resistance (HOMA-IR) at screening, day 1, and week 12. We assessed adverse events and concomitant drugs at each study visit. Outcomes The primary outcome was the absolute change in liver fat content from baseline to week 12. Responders were patients who achieved a clinically meaningful reduction of at least 5% in absolute liver fat content as measured by MRI-PDFF and assessed by the central radiology core facility at Duke University. Secondary outcomes included absolute and relative change in liver fat content C4 (Mayo Clinic, Rochester, MN, USA), normalisation of liver fat content (decrease to <5%) at week 12, and changes from 4 Published online March 5,

5 baseline and normalisation in ALT concentrations. Adverse events were assessed using the Common Terminology Criteria for Adverse Events (version 4.03; appendix p 4). Exploratory outcomes included changes in lipids and lipoprotein particles (LabCorp/LipoScience), as well as fibrosis biomarkers, including released N-terminal pro-peptide of type III collagen (pro-c3; Nordic Bioscience) and total enhanced liver fibrosis score (including the N-terminal pro-peptide of collagen III [PIIINP], the tissue inhibitor of metalloproteinase 1 [TIMP-1], and hyaluronic acid). Statistical analysis A minimal sample size of 75 patients was selected for the pre-study power calculation. Allowing for a 10% dropout rate, sample size calculations were based on a minimum of 22 patients per group. As shown by findings of previous studies, an absolute change in liver fat content of 5% is the minimal clinically relevant difference between active treatment and placebo, and correlates with improvements in liver histology. 17,22,23 Sample size power calculations were based on varying common standard deviations to achieve an overall (overall F test) significant treatment effect of 6% or higher reduction from baseline in liver fat content with 3 mg or 6 mg NGM282, and a reduction of 1% or lower in the placebo group (all at a significance level of 5%). The efficacy population was used to assess efficacy and pharmacodynamic endpoints, and included all randomly assigned and enrolled patients who received at least one dose (full or partial) of study drug and had a baseline and week 12 or end of treatment MRI-PDFF with at least one valid, non-missing post-dose efficacy or pharmacodynamic parameter value. We analysed efficacy with the intention-to-treat population defined as all patients who underwent randomisation. Patients in each treatment group were categorised as either responders or non-responders. The two patients in the NGM282 6 mg group who did not have an end of treatment MRI-PDFF imaging were imputed as non-responders. We compared the absolute change in liver fat content from baseline to week 12 between treatment groups using analysis of covariance (ANCOVA), with treatment group and diabetic status as cofactors and baseline liver fat content and ALT as a covariate at the 5% level of significance. Least-squares (LS) means with standard errors (SE), differences in LS means with SE, 95% confidence intervals (CIs), and the corresponding p values were calculated. Categorical endpoints were analysed using χ² test or Fisher s exact test. To adjust for multiple comparisons, the step down Bonferroni method was used for analysis between treatment groups. Normalisation was defined as patients with an MRI-PDFF value of 5% or lower at end of treatment. We analysed secondary and exploratory endpoints with the same methods as for the primary outcome. For continuous outcomes measured repeatedly over weeks, all values were summarised at each timepoint and comparisons to Placebo 3 mg NGM282 6 mg NGM282 (n=28) (Continued from previous page) Metabolic factors Glucose (mg/dl) (58 0) (35 7) (42 9) HbA1c 6 7% (1 4) 6 5% (1 0) 6 7% (1 2) Insulin (pmol/l) 4 0 (3 2) 5 3 (5 1) 3 6 (2 9) HOMA-IR 12 2 (22 1) 11 3 (14 7) 7 3 (5 8) Weight (kg) 97 8 (19 6) 95 5 (22 2) 98 2 (17 6) BMI (kg/m²) 35 6 (5 8) 34 0 (8 7) 34 7 (5 5) Waist circumference (cm) (12 3) (13 7) (17 9) Vital signs Systolic blood pressure (mm Hg) (17 8) (12 5) (13 0) Diastolic blood pressure (mm Hg) 82 6 (9 0) 77 0 (11 1) 79 5 (11 1) Comorbidities Type 2 diabetes 17 (63%) 15 (56%) 17 (61%) Hyperlipidaemia 8 (30%) 10 (37%) 15 (54%) Hypertension 21 (78%) 15 (56%) 18 (64%) Concomitant medications Antidiabetic Metformin 14 (52%) 13 (48%) 14 (50%) Insulin 4 (15%) 3 (11%) 5 (18%) GLP-1 agonists 3 (11%) 3 (11%) 5 (18%) DPP4 inhibitors 1 (%) 3 (11%) 0 SGLT2 inhibitors 3 (11%) 0 2 (7%) Other 4 (15%) 2 (7%) 4 (14%) Antilipidaemic Statins 7 (26%) 12 (44%) 14 (50%) Non-statins 8 (30%) 8 (30%) 8 (29%) Antihypertensives 25 (93%) 21 (78%) 23 (82%) Other Pioglitazone (4%) Vitamin E 2 (7%) 2 (7%) 0 Histopathology Fibrosis stage 1 11 (41%) 11 (41%) 10 (36%) 2 7 (26%) 7 (26%) 12 (43%) 3 9 (33%) 9 (33%) 6 (21%) Total NAS activity score 5 1 (1 1) 5 1 (1 0) 5 1 (0 9) Hepatocyte ballooning score 0 (none) (few ballooned cells) 13 (48%) 15 (56%) 21 (75%) 2 (many ballooned cells) 13 (48%) 12 (44%) 7 (25%) Steatosis score 0 (<5%) (5 33%) 6 (22%) 6 (22%) 3 (11%) 2 (34 66%) 14 (52%) 12 (44%) 15 (54%) 3 (>66%) 6 (22%) 9 (33%) 10 (36%) Lobular inflammation score 0 (none) (<2) 10 (37%) 12 (44%) 12 (43%) 2 (2-4) 14 (52%) 14 (52%) 16 (57%) 3 (>4) 2 (7%) 1 (4%) 0 (Table 1 continues on next page) Published online March 5,

6 Placebo 3 mg NGM282 6 mg NGM282 (n=28) (Continued from previous page) Liver fat content by MRI-PDFF Liver fat content (%) 16 8 (6 7) 18 1 (7 3) 19 5 (7 8) Fibrosis Pro-C3 (ng/ml) 18 9 (14 7) 19 2 (12 3) 14 9 (6 0) ELF score 9 9 (1 0) 9 7 (0 8) 9 6 (0 9) Hyaluronic acid (UG/L) 85 6 (90 3) 58 7 (47 4) 65 6 (67 6) PIIINP (UG/L) 11 8 (3 9) 12 5 (4 5) 10 7 (3 1) TIMP-1 (UG/L) (91 1) (65 8) (69 5) Data are mean (SD), n (%), unless stated otherwise. C4=serum 7-α-hydroxy-4-cholesten-3-one. DPP4=dipeptidyl peptidase 4. ELF=enhanced liver fibrosis. GLP-1=glucagon-like peptide 1 receptor agonists. HbA1C=glycated haemoglobin. HDL=high-density lipoprotein. HOMA-IR=homoeostasis model assessment estimated insulin resistance. INR=international normalised ratio. LDL=low-density lipoprotein. MRI-PDFF=magnetic resonance imaging-proton density fat fraction. NAFLD=non-alcoholic fatty liver disease. NAS=NAFLD activity score. PIIINP=N-terminal pro-peptide of collagen III. Pro-C3=N-terminal type III collagen pro-peptide. SGLT2=sodium-glucose cotransporter-2 inhibitors. TIMP-1=tissue inhibitor of metalloproteinase 1. One patient in the placebo group had total NAS score but not component values in the histopathology report. Table 1: Baseline demographics placebo were evaluated by ANCOVA. We used SAS (version 9.4) for all analyses. Safety and tolerability analyses were done with the safety population (all randomly assigned and enrolled patients who received at least one dose, full or partial, of study drug and had at least one post-dose safety evaluation). All safety endpoints were analysed descriptively. This trial is registered with ClinicalTrials.gov, number NCT , and was overseen by the funder s medical monitor and clinical research associate. Role of the funding source This study was designed by expert consultants in the non-alcoholic steatohepatitis field in conjunction with representatives of the funder. Data were collected by investigators, and managed, validated, and analysed by INC Research (Raleigh, NC, USA). The corresponding author and the funder had full access to all data in the study and had full responsibility for the decision to submit for publication. Results Between July 14, 2015, and Aug 30, 2016, 166 patients were screened, and 82 eligible patients were randomly assigned to receive 3 mg NGM282, 6 mg NGM282 (n=28), or placebo (n=27; figure 1). 82 patients were included in the efficacy analyses; two patients in the 6 mg dose group did not complete MRI-PDFF imaging for liver fat content assessment because of early withdrawal, and were imputed as non-responders. 82 patients were included in the safety analysis. Baseline demographics and disease characteristics of the three dosing groups were similar (table 1). At 12 weeks, 20 (74%) of 27 patients who received 3 mg NGM282 (relative risk [RR], 10 0; 95% CI ; p<0 0001), and 22 (79%) of 28 patients who received 6 mg NGM282 (11 4; 95% CI ; p<0 0001) met the primary endpoint versus only two (7%) of 27 in the placebo group (figure 2A). Reductions in absolute liver fat content were significantly greater in patients given 3 mg NGM282 (LS mean difference 8 8 [SE 1 4], 95% CI 11 6 to 6 1; p<0 0001) and 6 mg NGM282 ( 11 1 [1 4], 13 9 to 8 3; p<0 0001) compared with those given placebo (table 2, figure 2B). We recorded no significant differences between the NGM282 groups with regard to absolute liver fat content (p=0 112). Findings of post-hoc analysis showed that the greatest reductions in liver fat content from baseline to week 12 were in patients with baseline liver fat content higher than 20% as measured by MRI-PDFF (3 mg NGM282: 12 9 [SE 8 8], 95% CI 19 6 to 6 2]; p=0 002; 6 mg NGM282: 18 9 [6 3], 22 9 to 14 9; p<0 001; table 3). Patients who received either 3 mg or 6 mg NGM282 achieved greater relative reductions in liver fat content from baseline to week 12 than those in the placebo group (p< for both comparisons; figure 2C). Overall, 23 (85%) of 27 and 24 (86%) of 28 patients receiving 3 mg or 6 mg NGM282, respectively, achieved a clinically meaningful relative change in liver fat content ( 30% reduction; 3 mg NGM282: RR 11 5, 95% CI ; 6 mg NGM282: 12 5, ; p< for both comparisons). These changes have been shown to be associated with histological improvement in nonalcoholic steatohepatitis in previous studies. 22,24 Liver fat content completely normalised in seven (26%) patients in the 3 mg dose group and 11 (39%) in the 6 mg dose group; no patients receiving placebo achieved normalisation (table 2). Findings of post-hoc analysis showed that improve ments in liver fat content were consistent across a broad range of patient populations with no significant effect of baseline sex, race or ethnic origin, dia betes status, BMI, ALT concentration, fibrosis stage, or concomitant statin use (appendix p 10). We recorded greater reductions from baseline in ALT concentrations for both NGM282 groups (3 mg group: LS mean difference 35 1 [SE 6 0]; 95% CI 47 1 to 23 1; p<0 0001; 6 mg group: 36 5 [6 1]; 48 7 to 24 2; p<0 0001) at week 12 compared with placebo (table 3, figure 3, appendix p 11). Consequently, the relative decreases in ALT concentrations from baseline to week 12 were also significant in the 3 mg group and 6 mg group (3 mg group: 44 9 [SE 8 8], 95% CI 62 5 to 27 3; p<0 0001: 6 mg group: 47 4 [9 0], 65 4 to 29 5; p<0 0001) compared with placebo (appendix p 11). In patients with a baseline ALT concentration higher than 60 IU/L, significant reductions in ALT concentrations were recorded in patients given 3 mg NGM282 ( 56 [SD 42 8], 95% CI 84 8 to 27 3; p=0 0001) and 6 mg NGM282 ( 45 9 [23 2]; 95% CI 60 6 to 31 1; p<0 0001) by week 2. In 13 (24%) of 55 patients in the NGM282 groups, ALT concentrations achieved normalisation by week 2, and in 20 (36%) patients by week Published online March 5,

7 Similarly, treatment with NGM282 resulted in significant reductions in AST from baseline to week 12 compared with placebo (3 mg group: LS mean difference 20 4 [SE 4 7], 95% CI 29 8 to 11 0, p<0 0001; 6 mg group: 22 5 [5 0], 95% CI 32 4 to 12 7; p<0 0001; table 3, figure 3B, appendix p 5). There were no significant changes from baseline in alkaline phosphatase among the treatment groups (table 3, figure 3C). Of note, despite rebound, concentrations of ALT and AST remained significantly reduced at week 16 follow-up, 4 weeks after cessation of treatment (appendix). Re ductions in C4 concentrations from baseline to week 12 were significant for both NGM282 groups (p<0 0001); these changes were also significant when compared with the placebo group (p< for both comparisons; table 3, figure 3D). More than 65% of participants were at or below the limit of quantification for the C4 assay (0 9 ng/ml) as soon as week 1 of NGM282 treatment. Despite significant inhibition of the classic pathway of de novo bile acid synthesis as evidenced by C4 reduction, we noted no significant changes in serum vitamin D concentrations, an indicator of fat-soluble vitamin absorption, in patients given NGM282 (appendix p 6). At week 12, the absolute change in liver fat content was significantly correlated with change in ALT (Spearman correlation, r=0 46; p=0 0001), AST (r=0 37; p=0 0008), and C4 (r=0 53; p<0 0001) by post-hoc analysis (appendix p 7). At week 12, reductions in triglyceride concentrations were significantly greater in the NGM282 6 mg group (p=0 01) compared with the placebo group (table 3, appendix p 12). Low-density lipoprotein-c (LDL-C) concentrations were significantly increased in both the 3 mg NGM282 group and the 6 mg group (40 7 [8 5], ) compared with placebo (both p<0 0001). We noted no increase in LDL-C in the placebo group (table 3, appendix p 12). NGM282-associated LDL-C increase was driven mainly by an increase in large, buoyant LDL-C particles (appendix p 8), with concomitant reductions in the large triglyceriderich lipoprotein particles. Addition ally, we recorded no changes in high-density lipo protein (table 3, appendix p 13). Significant reductions in lipoprotein insulin resistance index and insulin resistance diabetes risk factor, which are lipid-based algorithms measuring hepatic and peripheral insulin resistance, 25 were also recorded in NGM282 groups (appendix p 8). The appendix (p 9) shows changes from baseline to week 12 in additional laboratory parameters. Assessments of non-invasive fibrosis biomarkers showed decreases after 12 weeks of NGM282 treatment. Pro-C3 and PIIINP are markers of collagen formation reflecting fibrogenic activity. Reductions in pro-c3 concentrations at week 12 were greater in participants in the 6 mg NGM282 group than the placebo group. Reductions in PIIINP concentrations from baseline to week 12 were 2 9 UG/L (p=0 02) and 1 5 UG/L Response (%) Absolute to relative change in LFC (%) A Placebo 3 mg 6 mg (n=28) C Figure 2: Treatment response (A), absolute change in liver fat content (B), relative change in liver fat content (C), and normalisation of liver fat content with 6 mg NGM282 (D), from baseline to week 12 Absolute and relative change values expressed as least squares means. LFC=liver fat content. MRI-PDFF=MRI proton density fat fraction. B 0 (p=0 1) for NGM282 3 mg and 6 mg, respectively, but these differences were not significant compared with placebo. TIMP-1 is a downstream marker of fibrogenic signalling. We noted significant reductions in TIMP-1 in the NGM282 3 mg group and 6 mg group (p=0 01) compared with the placebo group (p=0 02). We noted reductions in enhanced liver fibrosis (ELF) score from p< p=0 144 p< Placebo 3 mg 6 mg Placebo group 3 mg NGM282 group 6 mg NGM282 group (n=28) Absolute change in LFC (%) D 0 9 Baseline MRI-PDFF=24 1% 12 weeks MRI-PDFF=3 6% 3 mg NGM282 group vs placebo group p< p=0 112 p< Placebo 3 mg 6 mg p value 6 mg NGM282 group vs placebo group 40% 0% 40% 0% p value Liver fat content by MRI-PDFF Response 2 (7%) 20 (74%) 22 (79%) 10 0 ( ) < ( ) < Normalisation 0 7 (26%) 11 (39%) 30% decrease 2 (7%) 23 (85%) 24 (86%) 11 5 ( ) < ( ) < ALT Normalisation 1 (4%) 10 (37%) 10 (36%) 9 6 ( ) ( ) Data are n (%), relative risk (95% CI), unless otherwise stated. ALT=alanine aminotransferase. MRI-PDFF=MRI-proton density fat fraction. Table 2: Categorical outcomes at week 12 Published online March 5,

8 Change from baseline to week 12, placebo group Change from baseline to week 12, 3 mg NGM282 group Change from baseline to week 12, 6 mg NGM282 group (n=28) LS mean difference, 3 mg NGM282 group vs placebo group p value LS mean difference, 6 mg NGM282 group vs placebo group Liver fat content by MRI-PDFF Liver fat content (%) 0 1 (3 03) 9 6 (6 92) 12 5 (7 94) 8 8 (1 4; 11 6 to 6 1) < (1 4; 13 9 to 8 3) < Baseline 8 15% 0 22 (3 3) 5 43 (3 2) 6 28 (4 1) Baseline >15 20% 0 62 (2 5) 11 5 (3 0) 12 (2 4) Baseline >20% 0 3 (3 3) 12 9 (8 8) 18 9 (6 3) ALT Overall ALT (IU/L) 2 2 (21 1) 35 0 (44 6) 32 1 (35 3) 35 1 (6 0; 47 1 to 23 1) < (6 1; 48 7 to 24 2) < Baseline ALT >60 (IU/L) 9 5 (22 6) 63 1 (58 6) 54 5 (42 4) Baseline ALT (IU/L) 0 (18 9) 24 4 (7 3) 21 5 (17 2) Baseline ALT <40 (IU/L) 12 7 (22 0) 6 8 (12 5) 9 1 (8 6) 7α-hydroxy-4-cholesten-3-one (C4) C4 (ng/ml) 11 7 (32 4) 48 5 (57 1) 25 9 (26 2) 43 2 (7 4; 58 0 to 28 4) < (7 2; 53 6 to 25 0) < Liver function AST (IU/L) 0 6 (23 9) 17 9 (20 8) 17 0 (21 2) 20 4 (4 7; 29 8 to 11 0) < (5 0; 32 4 to 12 7) < ALP (IU/L) 0 1 (21 9) 0 5 (19 8) 2 3 (16 5) 0 7 (5 2; 9 6 to 11 0) (5 1; -8 4 to 11 9) 0 74 GGT (IU/L) 0 3 (32 7) 10 3 (62 3) 3 3 (53 3) 11 6 (13 7; 15 8 to 40 0) (14 0; 32 1 to 23 7) 0 77 Total bilirubin 0 (0 1) 0 (0 2) 0 (0 2) 0 1 (0; 0 to 0 2) (0; 0 to 0 2) 0 09 Albumin (g/l) 0 (0 3) 0 2 (0 4) 0 1 (0 3) 0 2 (0 1; 0 to 0 4) (0 1; 0 to 0 3) 0 12 Fibrosis biomarkers Pro-C3 (ng/ml) 1 21 (10 7) 5 4 (10 7) 3 6 (8 3) 4 0 (2 2; 8 3 to 0 3) (2 1; 8 9 to 0 4) Patients with >15% decrease 6/25 (24%) 17/23 (74%) 19/25 (76%) in pro-c3 ELF score 0 (0 5) 0 3 (0 6) 0 1 (0 6) 0 3 ( 0 2; 0 7 to 0) (0 2; 0 5 to 0 2) 0 33 Hyaluronic acid (UG/L) 7 0 (55 9) 5 5 (32 0) 22 6 (107 4) 7 5 (21 4; 50 5 to 35 4) (22 3; 23 to 66 3) 0 34 PIIINP (UG/L) 0 (3 1) 2 9 (5 6) 1 5 (4 3) 2 5 (1 3; 5 to 0) (1 3; 4 8 to 0 4) 0 10 TIMP-1 (UG/L) 5 2 (69 4) 25 9 (39 2) 18 0 (41 0) 40 7 (14 2; 69 1 to 12 3) (14 8; 64 8 to 5 4) Lipids Triglycerides (mmol/l) 0 1 (0 8) 0 4 (1 3) 0 5 (0 8) 0 1 (0 2; 0 5 to (0 2; 0 8 to 0 1) Cholesterol (mmol/l) 0 (0 6) 1 1 (1 0) 0 8 (0 8) 1 1 (0 2; 0 6 to 1 5) < (0 2; 0 4 to 1 2) HDL (mmol/l) 0 (0 2) 0 (0 3) 0 (0 3) 0 (0; 0 1 to 0 1) (0; 0 1 to 0 2) 0 57 LDL (mmol/l) 0 (0 5) 1 2 (1 0) 1 0 (0 8) 1 2 (0 2; 0 7 to 1 6) < (0 2; 0 6 to 1 5) < Concomitant statin 0 2 (0 5) 1 1 (0 7) 1 0 (0 6) No concomitant statin 0 1 (0 5) 1 2 (1 3) 1 0 (0 9) Metabolic factors Glucose (mg/dl) 1 7 (60 6) 2 9 (31 3) 2 9 (29 7) 4 1 (8 6; 21 3 to 13 2) (8 8; 17 6 to 17 4) 0 99 HbA1c (%) 0 1 (1 0) 0 2 (0 7) 0 1 (0 5) 0 2 (0 2; 0 5 to 0 2) (0 2; 0 4 to 0 3) 0 85 Insulin (pmol/l) 0 5 (2 0) 0 9 (4 9) 0 2 (2 3) 0 2 (0 8; -1 4 to 1 9 ) (0 9; -1 2 to 2 3) 0 50 HOMA-IR 1 13 (6 7) 0 9 (8 7) 0 7 (7 8) 1 4 (2 2; 5 8 to 2 9) (2 2; 5 to 3 9) 0 81 Weight (kg) 0 8 (2 4) 1 2 (3 5) 2 7 (3 4) 0 4 (0 8; 2 0 to 1 3) (0 9; 3 7 to 0 3) BMI (kg/m²) 0 2 (0 9) 0 4 (1 6) 0 9 (1 3) 0 3 (0 3; 0 9 to 0 4) (0 3; 1 5 to 0 1) Waist circumference (cm) 2 5 (12 8) 0 3 (4 5) 4 6 (12 6) 3 1 (2 9; 2 8 to 8 9) (3 0; 7 3 to 4 5) 0 65 Data are mean (SD), SE (95% CI), unless otherwise stated. ALP=alkaline phosphatase. ALT=alanine aminotransferase. AST=aspartate aminotransferase. ELF=enhanced liver fibrosis. GGT=α -glutamyl transpeptidase. HbA1C=glycated haemoglobin. HDL=high-density lipoprotein. HOMA-IR=homoeostasis model assessment-estimated insulin resistance. LDL=low-density lipoprotein. LS=least squares. MRI-PDFF=MRI-proton density fat fraction. PIIINP=N-terminal pro-peptide of collagen III. Pro-C3=N-terminal type III collagen pro-peptide. SD=standard deviation SE=standard error. TIMP-1=tissue inhibitor of metalloproteinase 1. Table 3: Continuous outcomes at week 12 p value baseline to week 12 in participants given 3 mg NGM282 compared with those in the placebo group (table 3). We recorded no significant placebo-adjusted changes from baseline to week 12 for the NGM282 groups in metabolic factors including glucose, HbA1c, insulin, HOMA-IR, or waist circumference (all p>0 05). Patients in the 6 mg NGM282 group reported significant reductions in weight and BMI compared with placebo 8 Published online March 5,

9 A Placebo 3 mg 6 mg B ALT (IU/L) End of treatment Follow-up AST (IU/L) End of treatment Follow-up C 80 D ALP (IU/L) End of treatment Follow-up Week C4 (ng/ml) End of 10 treatment Week Figure 3: Levels of alanine aminotransferase (A), aspartate aminotransferase (B), alkaline phosphatase (C), and 7α-hydroxy-4-cholesten-3-one (D) over time p< p<0 01. (table 3), but the reductions in liver fat content were independent of weight loss and BMI. Six (22%) of 27 patients in the 3 mg dose group and seven (25%) of 28 in the 6 mg dose group tested positive for antidrug antibodies; only three patients had titres above background (one patient from the NGM282 3 mg cohort had a titre of 1:22, two patients from the 6 mg cohort had titres of 1:18 and 1:23, respectively). Titres were low in the remaining ten patients (all with a titre of 1:3), which were considered background levels because a titre of 1:3 was also detected in two patients in the placebo group. Absolute liver fat content changed from 26 9% at baseline to 14 0% at end of treatment for the patient with a titre of 1:22 in the 3 mg cohort, from 15 0% to 7 0% for the patient with a titre of 1:18 in the 6 mg cohort, and from 14 0% to 18 0% for the patient with a titre of 1:23 in the 6 mg cohort. The NGM282 levels (C min ) were 16 6 ng/ml, 14 3 ng/ml, and 46 9 ng/ml at end of treatment in these three patients, respectively. None of the antidrug antibodies tested positive for neutralising activity against FGF19. Overall, 76 (93%) of 82 patients reported at least one adverse event, most of which were grade 1 (55 [67%]), and only five (6%) experienced adverse events that were grade 3 or worse. The most commonly ( 10%) reported adverse events were injection site reactions in 28 (34%) patients, diarrhoea in 27 (33%), abdominal pain in 15 (18%), and nausea in 14 (17%; table 4). These adverse events were reported more frequently in the NGM282 groups than the placebo group. Treatmentrelated adverse events occurred in 60 patients; the majority of which were mild (grade 1/2; table 4). Two (7%) patients in the 3 mg dose group had treatmentrelated grade 3 adverse events (one patient with aggravated abdominal pain, one patient with acute pancreatitis) compared with two (7%) in the 6 mg dose group (one patient with nausea or abdominal pain, one patient with depression), and one (4%) patient in the placebo group (increased hepatic enzyme). No grade 4 treatment-related adverse events were reported in any treatment group. One patient in the 3 mg dose group experienced a serious adverse event (acute pancreatitis), which was resolved with standard clinical support. This singular case of pancreatitis was unlikely to be related to study drug because the patient was on other medications known to be associated with pancreatitis and the patient also previously had a cholecystectomy. Study drug withdrawal because of adverse events occurred in three (11%) of 27 patients in the 3 mg NGM282 group (four events in three patients; aggravated abdominal pain, acute pancreatitis and increased lipase, aggravated nausea) and three (11%) of 28 patients in the 6 mg dose group (three events in three patients; loose stools, injection site reaction, and depression). No Published online March 5,

10 Placebo group 3 mg NGM282 group 6 mg NGM282 group (n=28) Most common adverse events (>10%) Injection site reactions 2 (7%) 11 (41%) 15 (54%) Diarrhoea 6 (22%) 11 (41%) 10 (36%) Abdominal pain 2 (7%) 8 (30%) 5 (18%) Nausea 1 (4%) 9 (33%) 4 (14%) Headache 5 (19%) 3 (11%) 5 (18%) Abdominal distension 1 (4%) 3 (11%) 4 (14%) Vomiting 0 2 (7%) 5 (18%) Frequent bowel 2 (7%) 3 (11%) 1 (4%) movements Increased appetite 0 2 (7%) 4 (14%) Constipation 1 (4%) 3 (11%) 1 (4%) Injection site bruising 3 (11%) 2 (7%) 0 Weight decreased (11%) At least one serious 0 1 (4%) 0 adverse event At least one adverse event leading to study drug temporary interruption or discontinuation 1 (4%) 5 (19%) 9 (32%) Treatment-related adverse events by severity grade Grade 1 14 (52%) 18 (67%) 23 (82%) Grade 2 4 (15%) 11 (41%) 12 (43%) Grade 3 1 (4%) 2 (7%) 2 (7%) Grade Acute pancreatitis. Table 4: Adverse events life-threatening events (grade 4) or patient deaths (grade 5) occurred during the study. Discussion In this randomised, placebo-controlled, double-blind study, NGM282 3 mg and 6 mg produced rapid and sustained improvements in liver fat content over 12 weeks of treatment. Greater treatment effect was recorded in patients with higher baseline liver fat content and ALT. In addition to changes in absolute liver fat content, NGM282 produced significant improvements in ALT, AST, and non-invasive serum fibrosis biomarkers (pro-c3 and ELF score). Decreases in liver fat content were correlated with decreases in C4, ALT, AST, and increases in LDL-C, consistent with a treatment effect on multiple pathways relevant to non-alcoholic steatohepatitis pathogenesis. The presence of antidrug antibodies in a small number of patients did not seem to have discernible effects on efficacy and safety outcomes, with none having neutralising activity against FGF19. NGM282 was generally well tolerated at both doses, with most treatment-related adverse events being mild in severity. Liver fat content quantification via MRI-PDFF has been previously shown to be sensitive in detecting liver fat changes and has been used in non-alcoholic steatohepatitis clinical trials for quantitative fat assessment. 26 Using paired MRI-PDFF and liver histology data, Patel and colleagues have shown that an absolute change in liver fat content of 4 1% with a relative change of 29 3% is associated with histological improvements of 2 or more point reduction in NAS (with 1 point each for steatosis and ballooning). 22 This suggests that change in liver fat may be a surrogate for early hepatocellular injury as well. Furthermore, weight-loss studies have demonstrated a clinically meaningful decrease of % in absolute liver fat content between the beginning and the end of the interventions. 23 Based on these considerations and in line with recent trials in the field, we have chosen a 5% reduction in absolute liver fat content as responder criteria for primary outcome, and a relative 30% reduction as a clinically meaningful change in liver fat content. NGM282-associated reductions in absolute liver fat content and relative liver fat content change exceed these criteria and thus are clinically important. Findings of preclinical animal studies have demonstrated that treatment with NGM282 leads to resolution of non-alcoholic steatohepatitis and improve ment in fibrosis through several mechanisms. In addition to suppressing the classic pathway of de novo bile acid synthesis, NGM282 inhibits fatty acid synthesis and de novo lipo genesis, presumably through mechanisms that eliminate or decrease hepatic bile acid toxicity and lipo toxicity, factors believed to be essential for the progression of non-alcoholic steatohepatitis to advanced liver diseases. 15 In view of the rapid and robust improvement in ALT and AST, in addition to the unparalleled reduction in liver fat content by a pharmacological agent in the current study, NGM282 could regulate multiple pathways to deliver weightindependent anti-steatotic, anti-inflammatory, and antifibrotic activities, and represent a potentially promising treatment for patients with non-alcoholic steatohepatitis. However, after cessation of NGM282 treatment, levels of transaminases increased toward baseline. Like FGF19, NGM282 potently suppresses expression of CYP7A1 via the FGFR4-βKlotho (KLB) receptor complex, reducing cholesterol catabolism. 16 Consistent with this mechanism of action, NGM282 significantly decreased serum C4 levels and produced significant increases in LDL-C, especially the large, buoyant LDL-C particles that are thought to be less atherogenic. The elevation in LDL-C has been previously shown in studies involving obeticholic acid, an agonist of farnesoid X receptor that induced FGF19 secretion. 27 Considering the increased cardio vascular risk associated with nonalcoholic steatohepatitis, it is important to explore the use of concomitant lipid-lowering therapies to support chronic administration of NGM282. Of note, treatment with SGLT-2 inhibitors for type 2 diabetes was associated with a beneficial impact on cardiovascular outcome despite significant increases in large particle LDL-C. 28 Further more, recent data have shown that concomitant 10 Published online March 5,

11 administration of a statin can mitigate NGM282-related increases in LDL-C. 29 A strength of this trial was the use of non-invasive MRI-PDFF to evaluate changes in liver fat content as opposed to histological evaluation. Subjects enrolled in this study had both pre- and post-treatment MRI exams, and MRI scans were read centrally in a manner similar to that described previously. 26 MRI-PDFF is highly sensitive to changes in liver fat content, is non-invasive and highly repeatable, provides quantitative estimates of overall liver fat content, and is less vulnerable to sampling error than liver biopsy alone. Importantly, MRI-PDFF has been shown to correlate well with improvements in liver histology. Both absolute change in liver fat content ( 4 1% reduction) and relative change in liver fat content ( 29 3% reduction) are associated with histological improvements of a 2 or more point reduction in NAS (with 1 point each for steatosis and ballooning, respectively), 22 and are more sensitive than histology-determined steatosis grade in quantifying liver fat changes. 24 Additionally, MRI examinations were randomised at final reading with regard to timepoint to reduce bias that might have resulted from knowledge of timepoints. Other strengths of this study include the correlation of measures of improvement with target engagement. Additionally, the robust treatment effect was observed across a broad range of patient populations with disease characteristics commonly seen in non-alcoholic steatohepatitis. Limitations of this study include a relatively short treatment period and small overall number of participants. Additional limitations include a lack of histology at the end of treatment, despite reductions in ALT, in addition to liver fat content, having been associated with histological improvement in non-alcoholic fatty liver disease activity score and fibrosis scores. 30 Furthermore, the selection of 5% reduction in absolute liver fat content to define responders might have underestimated treatment effect, as patients with lower baseline liver fat content are less likely to achieve this endpoint. Another limitation was that relatively few non-white participants were included, and numerically more men (non-significant) were included in the NGM282 groups, although both race and sex had no effect on reductions in liver fat content in post-hoc analysis. Future trials should include longer term studies in larger cohorts of non-alcoholic steatohepatitis patients with comparative histological assessment. In view of the similar efficacy of NGM282 3 mg and 6 mg in lowering liver fat content, doses lower than 3 mg should be evaluated to better characterise the efficacy and tolerability profile. Future research may examine potential steatosisindependent, anti-inflammatory mechanism of NGM282. These studies may have important implications on clinical practice for the treatment of a chronic disease such as non-alcoholic steatohepatitis. In conclusion, 3 mg and 6 mg doses of NGM282 significantly and rapidly (over 12 weeks) reduced liver fat content, as well as markers of liver inflammation and fibrosis. Additionally, both doses were generally well tolerated. These results support the further exploration of NGM282 for the safe and effective treatment of non-alcoholic steatohepatitis. Contributors SAH, MER, MFA, MK, SJR, AMD, and RL participated in study design. SAH, MER, MFA, JFT, AHP, HLA, MK, MRB, RL were responsible for data collection. SAH, MJJ, LL, SJR participated in data analysis. SAH, MER, MFA, JFT, AHP, HLA, MK, MRB, LL, SJR, AMD, RL participated in data interpretation. All authors participated in manuscript review and writing. SAH, LL, SJR were responsible for preparation of the tables and figures. Declaration of interests SAH reports grant and research support from Allergan, Conatus, Galectin, Galmed, Genfit, Gilead, Immuron, Intercept, Madrigal, NGM Bio, and TaiwanJ; participates on Speakers Bureaus for Alexion and Gilead; is a consulting adviser for Allergan, Chronic Liver Disease Foundation, Cirius, Echosens, Fibrogen, Galmed, Genfit, Gilead, Intercept, Madrigal, NGM Bio, Novartis, Perspectum, and Pfizer. MER reports being on advisory committees or review panels for Intercept, Abbvie; consulting for Intercept, Novartis, NGM Bio, Shire, and Fibrogen. MFA reports grant and research support from NIH/NIDDK, Allergan, Conatus, Galectin, Galmed, Genfit, Gilead, Immuron, Intercept, Madrigal, Shire, NGM Bio, BMS, Excelenz, Boehringer Ingelheim, Prometheus and TaiwanJ; is on Speakers Bureaus for Alexion; is a consulting adviser for Allergan, TaiwanJ, NGM Bio, BMS, and Pfizer. MRB reports grant and research support from NGM Bio, TaiwanJ, Madrigal, Siemens Healthineers, General Electric Healthcare, non-alcoholic steatohepatitis Clinical Research Network; is a consulting adviser for RadMD. MJJ reports consulting for NGM Bio. SJR, LL, and AMD report being employees and stockholders of NGM Bio. RL reports grant and research support from Daiichi Sankyo, AGA, Merck, Promedior, Kinemed, Immuron, Adheron; is on advisory committees or review panels for Galmed, Tobira, Arrowhead; consulting for Gilead, Corgenix, Janssen, Zafgen, Celgene, Alnylam, Enanta, NGM Bio and Deutrx. All other authors declare no competing interests. Acknowledgments We thank all of the patients who participated in this study, and the investigators, study coordinators, and staff of all of the participating clinical centres for their support and assistance; Heather King and Claudette Knight of Percolation Communications, LLC for medical editorial assistance. Medical editorial assistance was funded by NGM Biopharmaceuticals, Inc. References 1 Rinella ME. Nonalcoholic fatty liver disease: a systematic review. JAMA 2015; 313: Starley BQ, Calcagno CJ, Harrison SA. Nonalcoholic fatty liver disease and hepatocellular carcinoma: a weighty connection. Hepatology 2010; 51: Wong RJ, Aguilar M, Cheung R, et al. Nonalcoholic steatohepatitis is the second leading etiology of liver disease among adults awaiting liver transplantation in the United States. Gastroenterology 2015; 148: Wong VW, Chitturi S, Wong GL, Yu J, Chan HL, Farrell GC. Pathogenesis and novel treatment options for non-alcoholic steatohepatitis. Lancet Gastroenterol Hepatol 2016; 1: Arab JP, Karpen SJ, Dawson PA, Arrese M, Trauner M. Bile acids and nonalcoholic fatty liver disease: molecular insights and therapeutic perspectives. Hepatology 2017; 65: Puri P, Daita K, Joyce A, et al. The presence and severity of nonalcoholic steatohepatitis is associated with specific changes in circulating bile acids. Hepatology 2017; published online July 11. DOI: /hep Mouzaki M, Wang AY, Bandsma R, et al. Bile acids and dysbiosis in non-alcoholic fatty liver disease. PLoS One 2016; 11: e Kliewer SA, Mangelsdorf DJ. Bile acids as hormones: the FXR-FGF15/19 pathway. Dig Dis 2015; 33: Published online March 5,