AB Science webcast of May 30, 2024

AB Science Web Conference Development Update 30 May 2024 2 Disclaimer This presentation, together with the material set forth herein, does not constitute an offer of securities for sale nor the solicitation of an offer to purchase securities in any jurisdiction. Distribution of such presentation in certain jurisdiction may constitute a breach of applicable laws and regulation. This document is solely for your information on a confidential basis and may not be reproduced, redistributed or sent, in whole or in part, to any other person, including by email or by any other means of electronic communication. In particular, neither this document nor any copy of it may be taken, transmitted or distributed, directly or indirectly, in the United States, Canada, Japan or Australia. The distribution of this document in other jurisdictions may be restricted by law and persons into whose possession this document comes should make themselves aware of the existence of, and observe, any such restrictions. Neither the Company, nor any of its advisors and representatives may accept any responsibility for any loss or damage incurred by the use of this document or the information set forth herein. Neither the Company, nor any of its advisors and representatives takes any undertaking nor guarantees, whether explicitly or tacitly, the accuracy or the completeness of the information set forth herein. Neither this document, nor any part of it, shall form the basis of, or be relied upon in connection with, any contract or commitment whatsoever. In particular, in France, any decision to purchase such securities shall rely solely on the documents that have been reviewed by the Autorité des Marchés Financiers (the “AMF”) and/or published by the Company. This document does not constitute an offer to purchase any financial instruments in the United States. Securities mentioned in this document have not been and will not be registered under the Securities Act of 1933, as amended (the “Securities Act”) and may not be offered or sold in the United States absent registration or an exemption from the registration requirements of the Securities Act. The Company does not intend to register any offering in all or in part or to make a public offer of securities in the United States. This document contains information on the objectives of the Company along with some projections and forward-looking statements. The reader’s attention is drawn to the fact that these objectives may not be fulfilled, and the forecasts or information provided may prove erroneous, and the Company is not required to update such information. Past performance is no guide to future performance and persons needing advice should consult an independent financial adviser. CONDITIONAL APPROVAL OF MASITINIB IN ALS WITH EMA The CHMP confirmed that the safety of masitinib is deemed acceptable 4 CHMP Latest Assessment Report “Having considered the data from masitinib studies, the safety profile of mentioned medicinal product is considered acceptable”;Conclusion on clinical safety However, the CHMP was not able to conclude on a favorable benefit due to the following methodological issues 5 Significant effect was not demonstrated in the total study population (Normal + Fast progressors) and exclusion of fast progressors from the primary population is not supported Primary population without fast progressors Handling of Missing Data with Jump to Reference (J2R) Imputation of a penalty for missing data (jump to reference) should be applied to all discontinuations occurring in the masitinib treatment arm New target population is identified post-hocSubgroup identified Topic Objection GCP Issues Identified GCP deviations to the protocol cannot be corrected retrospectively# 1 # 2 # 3 # 4 As per guideline, impact analyses of all findings that could not be corrected were performed and showed no impact Resolution of Issue #1 – GCP Issues 6 Inspection findings which are likely to influence the benefit-risk evaluation • Deficiencies in blinding of study medication • Deficiencies in randomization • Violation of diagnostic inclusion- and exclusion criteria • Missing source documentation • Faults in data management, statistical programming and analyses • Fraud and other scientific misconduct No deficiencies • Violation of procedures related to the assessment of the primary efficacy endpoint Analysis which does not identify any impact on efficacy • Inadequate reporting of adverse events and other safety endpoints Safety : corrected and processes improved Inspection findings which may influence the benefit-risk evaluation • Violation of study procedures regarding rescue medication • Inadequate calibration of instruments, equipment etc. related to efficacy assessment • Rounding issues • Failure to document pre-specification of analyses prior to breaking study blind • Deficiencies in storage of study medication No deficiencies • Violation of inclusion and exclusion criteria (other than diagnostic criteria) Analysis which does not identify any i mpact on efficacy/safety • Deviations from protocol-specified visit windows Rules defined by DSMB shows no impact • Discrepancies between the clinical study report and the actual conduct of the study New CSR Version was released w ith final actual data • Deficiencies in preparation and administration of study medication at investigational sites One underdosed error in 42% of study patients. No impact Guidance (EMA/868942/2011) Framework Inspections findings Action taken As per guideline, the benefit-risk remains evaluable because critical data from study AB10015 are reliable Resolution of Issue #1 – GCP Issues 7 Safety data were deemed acceptable Primary efficacy data are reliable ▪ Eligibility criteria, riluzole dosing, masitinib dosing, visit windows were investigated ▪ Impact analysis showed no impact of these deviations, favouring masitinib or modifying study outcome ▪ Raters were experienced in ALSFRS-R scale according to inspectors ▪ Inspection recorded few discrepancies in ALSFRS-R, which proves it was not systematic o Findings in 3 (1.8%) out of 168 patients inspected (43% of study) o Finding in 1 out of 4 sites inspected* ▪ Nevertheless, the ALSFRS-R data have been reassessed at source by independent certified raters and there was no impact ▪ The safety profile of masitinib is already considered acceptable by the CHMP Deviations on other study data have no impact ▪ Guidance recommendations are fulfilled (1) o In superiority studies, […], inspection findings merely indicating increased variability and not introducing bias favouring one treatment over the other are relatively unproblematic in the interpretation of the study results o It is important to assess whether the findings affect the interpretation of the primary efficacy endpoint or important safety endpoints ▪ Benefit risk can be evaluated (1) 19 September 2012 EMA/868942/2011. Good Clinical Practice Inspectors Working Group (GCP IWG). Chapter 2. General Notes on Findings There is a necessity in ALS drug development to use a more homogenous population with greater chance to reach week 48 and minimize missing data, justifying the exclusion of Fast Progressors Resolution of Issue #2 - Primary population without fast progressors 8 ▪ Clinical relevance of the predefined 1.1 points/month cut-off for post-onset ΔFS was based on data published prior to initiation of study AB10015 [1] ▪ ΔFS has been recognized in many publications as a robust prognostic factor for survival [2] ▪ Importantly, post-onset (early) ΔFS has been shown to be the clinical determinant of greatest prognostic importance, while other calculations of ΔFS (post- randomization) showed no prognostic value [Requardt 2021]. ▪ The ΔFS from onset of symptoms has been widely used as a selection criteria in recent clinical trial, including for pivotal studies of edaravone and tofersen ▪ The ΔFS from onset of symptoms is now recognized by the community as a common tool used in clinical practice [3] ▪ The risk of patient misclassification due to recollection error of first symptom onset is low according to AB10015 data (recall of month of symptom onset is sufficient for correct classification of 98% of patients) and study was successful from a cut of 0.8 to 1.4 ▪ Excluding the Fast Progressors from primary analysis was necessary o In pivotal studies, there is a need to select a more homogeneous population o For a 48-week study in ALS, there is a need to minimize missing data via patient selection criteria Patient Status (ΔFS) Placebo M4.5 Normal 26.5% 30.5% Fast 52.6% 56.5% AB10015 – Discontinuation rate at week 48 according to patient status [1] [Kollewe K, J Neurol Sci 2008 ; [2] [Kollewe K, J Neurol Sci 2008 ; Hannaford A, Muscle Nerve. 2023 ; Requardt MV, J Clin Med. 2021 ; Kjældgaard AL, BMC Neurol. 2021 ; Su WM, EBioMedicine. 2021 ; Labra J, J Neurol Neurosurg Psychiatry. 2016 ; Kimura F, Neurology 2006 [3] Ludolph A; Muscle Nerve. 2024 Primary analysis: The definition of discontinuation reasons on which to apply a penalty (called jump to reference) is conventional and alternative methods recommended by EMA were positive Resolution of Issue #3 - Handling of Missing Data with J2R 9 ▪ Jump to Reference (J2R) was not narrow but conventional o J2R was conventional with penalty applied to discontinuations for toxicity and lack of efficacy, and was positive (p=0,0389) o J2R applied to discontinuations for toxicity, lack of efficacy and travel was positive (p=0,0376) o J2R for all discontinuations is extreme [Carpenter 2013] and close to statistical significance (p=0.0678) Analysis Difference, [95% CI] p-value J2R Lack of Efficacy/ any AE 2.8 [0.15; 5.46] 0.0389 J2R Lack of Efficacy/ any AE/Travel 2.82 [0.16; 5.48] 0.0376 J2R All discontinuations 2.31 [-0.17; 4.80] 0.0678 Tipping-Point Analysis = 76% 2.48 [0.00; 4.96] 0.0498 CIR 2.67 [0.03; 5.32] 0.0477 ▪ Second alternative method recommended by EMA was positive : Copy Increment in Reference (CIR) o Was originally recommended by the CHMP (EMEA/H/C/4398, D150 JAR) o Was considered acceptable in other ALS EPAR (EMA/CHMP/487533/2023) o positive (p=0.0477) ▪ Reasons for discontinuation were documented in the eCRF and classification MNAR versus MAR was prespecified in the SAP before unblinding ▪ First alternative method recommended by EMA was positive : Tipping-point for penalty on all discontinuations o 76%, meaning that 24% of masitinib effect is enough to become significant o 76% is “within what would be empirically expected” according to Rapporteur (EMEA/H/C/4398, D180 JAR) There was a benefit on quality of life, as measured by ALSAQ-40 based on a non LOCF method and a trend of benefit on CAFS, although the study was not powered to demonstrate efficacy on CAFS Resolution of Issue #3 - Handling of Missing Data with J2R 10 Masitinib 4.5 mg/kg/day – Normal Progressors - ALSAQ-40 Analysis Difference, [95% CI] p-value Multiple Imputation -7.27 [-13.18; -1.37] 0.0161 Multiple Imputation with J2R -5.99 [-11.55;-0.44] 0.0343 CIR -6.04 [-11.51;-0.57] 0.0305 CAFS is close to significance with p-value = 0.07, even though the study was not designed for CAFS as a primary endpoint Masitinib 4.5 mg/kg/day – Normal Progressors - CAFS analysis Treatment group N Mean Score Difference of means p-value Placebo 113 100.77 14.95 0.0776 Masitinib 4.5 105 115.72 Significant ALSAQ-40 benefit based on conservative imputation methods Time to event PFS or EFS, accounting for tracheostomy ventilation and gastrostomy, showed a significant benefit in favour of masitinib Resolution of Issue #3 - Handling of Missing Data with J2R 11 Progression Free Survival was prespecified There was a significant PFS benefit (+4 months) with a p=0.0159 Masitinib 4.5 mg/kg/day – Normal Progressors. PFS analysis PFS is robust since all tracheostomy and most of the ventilation or gastrostomy events occurred after progression Summary of interventions with respect to time of progression Treatment group N Median months [95% CI] Wilcoxon p- value Placebo 113 16 [11; 19] 0.0159 Masitinib 4.5 105 20 [14; 30] Event Type Treatment Total(n) Before Progression (%) After Progression (%) Tracheostomy Placebo (N=114) 5 0% 100.0% Masitinib 4.5 (N=106) 11 0% 100.0% Ventilator Placebo (N=114) 21 0% 100.0% Masitinib 4.5 (N=106) 24 8.3% 91.7% Gastrostomy Placebo (N=114) 20 15.0% 85.0% Masitinib 4.5 (N=106) 22 27.3% 72.7% Event Free Survival (EFS), defined as the earliest between progression, tracheostomy, ventilation, gastrostomy, or death event, remained statistically significant (p=0.01), showing a median benefit of +3 months Masitinib 4.5 mg/kg/day – Normal Progressors. Analysis of PTVGFS Treatment group N Median months [95% CI] Wilcoxon p-value Placebo 113 11 [10; 17] 0.0162 Masitinib 4.5 105 14 [12; 20] PTVGFS: Progression Tracheostomy Ventilation Gastrostomy Free Survival There was a trend of +6 months increase in median overall survival (OS) Resolution of Issue #3 - Handling of Missing Data with J2R 12 As requested by the Rapporteur, new long-term OS analysis was performed, censoring placebo patients at the time they switched to masitinib treatment in the NPP program This retreatment is essential because OS was significantly improved for placebo patients who switched to masitinib 4.5 mg/kg/day as compared with placebo patients who did not switch Analysis of overall survival in the NPP program (placebo switch vs no switch to masitinib) With the above mentioned retreatment, there was a trend in overall survival improvement (+6 months) with a p-value of 0.0761 Masitinib 4.5 mg/kg/day – Normal Progressors. OS analysis from baseline (censoring of placebo patients at time of switch to masitinib - ITT) Treatment group N Median months [95% CI] Wilcoxon p-value Placebo who did not switch 53 62 [49; NE] 0.0367Placebo who switched to Masitinib 4.5 25 69 [44; NE] Treatment group N Median months [95% CI] Wilcoxon p-value Placebo 114 40 [30; 49] 0.0761 Masitinib 4.5 106 46 [33; 69] EMA guidance (EMA/CHMP/539146/2013) on subgroup applies even to post-hoc analyses Resolution of Issue #4 – Subgroup identified 13 Scenario 2 of the guidance is applicable because there is a bias in a subset of patients EMA guidance applies to post- hoc analyses initiated by the Applicant “[…] it is of interest to identify post-hoc a subgroup, where efficacy and risk-benefit is convincing”. subgroup analysis applies “ […] irrespective of whether it is the company or the regulator that is specifying additional investigations of interest”. ▪ The clinical data are statistically persuasive in the primary analysis population ▪ But there are risks and uncertainties present in a subset of the population, because there is a bias in this subset ▪ Therefore, it might be of interest to identify a subgroup that has not been pre-specified as part of the confirmatory testing strategy, where efficacy and risk-benefit would be convincing There was a disbalance in a subset of the study population (patients with complete loss of function in at least one individual component of the ALSFRS-R) because ALSFRS-R score was minimized but not stratified by category of severity Resolution of Issue #4 – Subgroup identified 14 Distribution of patients with complete loss of function in at least one individual component of the ALSFRS-R in the primary analysis population Normal Progressor Placebo N=113 Masitinib 4.5 mg/kg/d N=105 ALS with complete loss of function in any individual component of the ALSFRS-R (score of zero on any item) 8.0% 20.0% A greater proportion of patients with any complete loss of function were randomized in the masitinib arm (20%) as compared with the control arm (8%) The disease severity in patients with any complete loss of function was higher in the masitinib arm Population Stat Placebo Masitinib 4.5 mg/kg/d 1 item with score 0 n [Q] 8 [8] 10 [10] 2 items with score 0 n [Q] 0 7 [10] 3 items with score 0 n [Q] 0 3 [9] 4 items with score 0 n [Q] 1 [4] 1 [4] Total n [Q] 9 [12] 21 [33] The bracketed [Q] represents the number of items with a 0-score. Some patients (n) had a 0-score (Q) in more than one item The treatment effect in the subgroup “prior to any complete loss of function” is exceptionally strong, and remains strong when considering both Normal and Fast Progressors Resolution of Issue #4 – Subgroup identified 15 Differential treatment effect (M4.5 vs placebo) Primary analysis population (Normal Progressors) Subgroup ‘Prior to any complete loss of function’ (Normal Progressors) Subgroup ‘Prior to any complete loss of function’ (Normal + Fast Progressors) ∆ALSFRS-R (CIR) Diff. of mean 2.68 3.13 2.76 p-value 0.0462 0.0308 0.0538 CAFS Relative benefit 14.8% 20.2% 18.4% P-value 0.0776 0.0290 0.0346 ALSAQ-40 (CIR) Diff. of mean -6.04 -6.22 -6.74 p-value 0.0305 0.044 0.0235 FVC (CIR) Diff. of mean 5.85 7.59 7.79 p-value 0.0931 0.0384 0.0369 Median PFS Gain + 4 months + 9 months + 5 months Median [95% CI] 20 [14; 30] vs 16 [11; 19] 25 [17; NE] vs 16 [11; 19] 20 [14; 30] vs 15 [11; 19] p-value log rank 0.0159 0.0057 0.0183 Median OS (Long-term) (censoring of placebo at time of switch to masitinib) Gain + 6 months + 12 months + 8 months Median [95% CI] 46 [33; 69] vs 40 [30; 49] 53 [36; NE] vs 41 [30; 54] 46 [30; 69] vs 38 [29; 49] p-value log rank 0.0761 0.0192 0.0684 ▪ Clinical data from the subgroup are exceptionally strong due to the consistency and magnitude of effect across endpoints ▪ 12 months survival benefit cannot be disregarded Next step 16 AB Science will liaise with the EMA to define the appropriate pathway to registration ▪ AB Science has the option to request a re-examination ▪ Re-examination is assessed by new Rapporteur and new Co-Rapporteur ▪ The re-examination provides the possibility of a Scientific Advisory Group) that can give recommendations on key points ▪ Key points could include o Application of EMA guideline on the GCP o Application of EMA guideline on subgroup o Application of the two recommendations for handling of missing data o Whether excluding Fast progressors from primary analysis is justified CONDITIONAL APPROVAL OF MASITINIB IN ALS WITH HEALTH CANADA Clinical objections raised by Health Canada were slightly different from the ones raised by EMA Health Canada Objections 18 Resolved Issue ▪ Changing from phase 2 to phase 3 ▪ Amendments may be an inevitable and were not data-driven ▪ There may be a need to limit heterogeneity ▪ Post-onset decline of 1.1 point per month may be relevant Primary analysis population Argumentation ▪ Amendment entirely blinded and sufficiently early o Implemented 2.5 years prior to study completion o Only 8 Fast progressors across 3 treatment arms could have reached the study’s primary analysis timepoint o When removing the 12% of Normal progressors that could have reached the study timepoint, the study remains positive o Study was positive at interim analysis in Normal + Fast ▪ Need to minimize expected high missing data due to discontinuations from Fast progressors with a long time-point of 48-weeks Pending Issue ▪ Amendments were late and not sufficiently justified Resolved Issue ▪ The non-linearity of the distribution of the ALSFRS-R data of the primary analysis (ANCOVA test) was resolved by the positive pre-specified rerandomization test Statistical method for imputation of missing data in primary analysis Argumentation ▪ Sensitivity analysis of the primary analysis based on non LOCF recognized methods are successful and convergent ▪ CAFS endpoint, incorrectly assumed by the Agency to be based on LOCF methodology, whereas it is not, approached the conventionally statistically significant outcome of 5% (p=0.0776), even though the study was not powered for this secondary endpoint. Pending Issue ▪ Missing data have been treated in the planned primary and sensitivity analyses with LOCF, potentially creating a bias in favor of treatment Subgroup analysis Argumentation ▪ EMA guidance applies to post-hoc analyses initiated by the Applicant “[…] it is of interest to identify post-hoc a subgroup, where efficacy and risk-benefit is convincing”. Pending Issue ▪ New proposed claim in patients prior to any loss of function is considered post hoc Available preclinical and clinical data in the context of a full approval Conditional approval raises the question of an approval with a single study that requires very compelling evidences. Available data from masitinib program are robust and will support full approval, provided a confirmatory study is positive 20 ▪ Masitinib has a validated mechanism of action, targeting the innate immune system, via modulation of mast cells and microglia ▪ Masitinib exerts a protective effect on the central nervous system o Reduction of microgliosis and aberrant glial cells through CSF-1R inhibition o Masitinib prevents motor neuron degeneration o Masitinib treatment significantly prolonged survival in post-paralytic SOD1G93A rats ▪ Masitinib exerts a protective effect on the peripheral nervous system o Mast cells are immune cells that regulate inflammation. Upon activation they release many proinflammatory mediators. o Mast cell infiltration and degranulation contribute to neuromuscular pathology in post-paralytic SOD1G93A rats o Masitinib-induced mast cell inhibition significantly reduced rate of neuromuscular junction (NMJ) and motor deficits ▪ Masitinib has demonstrated ability to lower blood levels of neurofilament light (NfL) in a neurodegenerative disease model (EAE model) Differential treatment effect (M4.5 vs placebo) Subgroup ‘Prior to any complete loss of function’ (Normal Progressors) ∆ALSFRS-R (CIR Analysis) Diff. of mean 3.13 p-value 0.0308 CAFS Relative benefit 20.2% P-value 0.0290 ALSAQ-40 (CIR Analysis) Diff. of mean -6.22 p-value 0.044 FVC (CIR Analysis) Diff. of mean 7.59 p-value 0.0384 Median PFS Gain + 9 months Median [95% CI] 25 [17; NE] vs 16 [11; 19] p-value log rank 0.0057 Median OS (Long-term) (censoring of placebo at time of switch to masitinib) Gain + 12 months Median [95% CI] 53 [36; NE] vs 41 [30; 54] p-value log rank 0.0192 ▪ Study AB10015 is a 48-week study, which is a stronger evidence than a 24-week study, and is same time point as the confirmatory study ▪ Study AB10015 demonstrated a significant treatment effect ▪ Data are exceptionally strong in the population close to the confirmatory study and that could be the claim o Significant benefit on functional score o Significant benefit on quality of life o Significant benefit on long-term overall survival Preclinical data Clinical data

The release as fetched from its publisher. ab-science.com ↗