Discussion on clinical safety

Exposure

The assessment of AZD1222 safety is based on the interim analysis of the results from all studies pooled in the total Safety analysis Set, comprising 23,745 participants (12,021 subjects: any dose of AZD1222, 11,724: control vaccine or placebo) from four individual studies, COV001, COV002, COV003 and COV005.

Slight differences regarding the methodology for collection of AEs and the measurement of the severity scale between trials were observed. Due to differences between the methods applied in COV005 compared with the other studies, information relating to solicited AEs from COV005 should not be pooled with the other studies.

Reactogenicity were collected in a subset of 2,648 participants receiving Dose 1 SD for 7 days following each vaccination, and 553 receiving dose 1 of LD. As 986 participants were enrolled in COV005, these were excluded from the pooled reactogenicity subset.

Among the 12,021 participants, dose 1 SD was given to 10,069 subjects and dose 1 LD to 1,947 subjects. A two-dose study intervention regimen was received by approximately two-thirds of participants. In the AZD1222 group, most participants had received two doses of the SD/SD regimen (54.6%) or a single SD before the data cut-off (28.7%).

In the Any Dose for Safety Analysis set, the median number of days of follow up was 105 days in the

AZD1222 treatment group and 104 days in the control group from the first dose. The median exposure in the Dose 1 SD Safety analysis set was 90 days in the AZD1222 group and 89 days in the control group. The maximum duration of follow up was 196 days from the first dose. The median of duration of follow up from the second dose was 55.6 days in AZD1222 and 54.7 days in control group (Any dose for Safety Analysis Set).

Demographic and baseline characteristics were generally similar among participants who received AZD1222 and the control treatment in the Any dose Safety analysis. 8.9% of participants were aged 65 or older. Most participants (95.1%) were seronegative at baseline. Approximately one-third of participants had at least one comorbidity at baseline (35.8%). The demographic characteristics (age, sex, race) are similar for the any dose and the Dose 1 SD safety analysis set.

HIV participants were included in COV002 and COV003 studies but excluded from the pooled analysis. A safety analysis of HIV population is lacking; therefore it is included as missing information in the RMP.

Adverse events

Solicited AEs, unsolicited AEs, SAEs (including deaths) and AESIs were evaluated.

Solicited Adverse Events: Any local and systemic AE were reported more frequently in AZD1222 than in control group. Solicited local and systemic AEs were generally milder and reported less frequently after the second dose than after first dose of AZD1222. By dose interval, the reactogenicity of AZD1222 was lower in participants in the < 6 weeks dosing interval compared with participants in the > 6 weeks dosing intervals, however this may have been confounded by differences in the population studied or other factors. Thus, interpretation of an effect due to the dose interval should be undertaken with caution.

The most frequently reported solicited local AEs in AZD1222 group were tenderness, followed by pain. The most frequently reported solicited systemic AEs in AZD1222 group were fatigue and headache, followed by muscle pain, malaise, feverishness, chills, joint pain and nausea.

Unsolicited Adverse events: Any unsolicited AEs were reported more frequently in AZD1222 group than in control treatment and generally reflected reactions to vaccination such as vaccination site pain, headache, pyrexia and myalgia. A majority of events was mild to moderate in severity, showing a reduction of the percentages (related or not) after the second dose in both the study vaccine and the comparator. The most frequently reported unsolicited AEs predominantly occurred within ≤7 days of any dose. There were no unsolicited AEs reported by preferred term in more than 2% of subjects within 8-28 days after any dose either AZD1222 or control group.

A noticeable imbalance in the frequency of unsolicited AEs in the Nervous System Disorder class between the AZD1222 and the control group is observed in the pooled results for the any dose safety analysis set.

Further, the imbalance is also present in the reported unsolicited AEs related to the ADZ1222 vaccine.

There were 3 cases of facial paralysis in the AZD1222 group and 3 in the control group. For one of the cases in the AZD1222 group, causality to the vaccine could not be excluded. There was no imbalance between the study groups in the occurrence of Bell’s palsy. No risk is identified as only a single case occurred for which

causality to the vaccine cannot be determined. Facial paralysis is however considered an adverse event of special interest (AESI) and will be characterized in the planned PASS studies and other ongoing clinical studies. A targeted questionnaire will be utilized for adverse reaction follow-up as a part of routine pharmacovigilance activities post-authorisation.

AESIs: The overall incidence of AESIs was low: 0.8 % of participants in the AZD1222 group (95 cases) and 1.1 % in the control group (126 cases). The majority of the reported events were paraesthesia, hypoesthesia and muscular weakness that account for 57 of 95 AESIs in the AZD1222 and 76 of 126 cases in the control group. There were 30 participants (0.2%) in the AZD1222 group and 44 (0.4%) in the control group who reported AESIs considered related by the investigator.

There was no imbalance of paraesthesia, dysesthesia, Hypoesthesia and Hyperaesthesia between the two groups, and the numbers were limited. The causal relationship with AZD1222 could not be established.

Three AESIs in total were reported as SAEs across treatment groups, of which transverse myelitis and myelitis were considered possibly related to the intervention by the investigator although causation could not be established, and multiple sclerosis was considered unlikely related to the intervention. The SAEs facial spasm and migraine, belonging to the same CNS SOC, may be potentially related to the intervention.

In the US study DC8110C00001 an additional two events which were AESIs and serious were reported: an event of sensory neuropathy and an event of Chronic Inflammatory Demyelinating Polyradiculopathy (CIDP).

Further, a case of acute encephalopathy was reported in the COVISHIELD study, which is a suspected nutritional encephalopathy, but other aetiologies have not been ruled out.

It is uncertain whether the study treatment was the cause of any of these events.

Based on the reported neurological AEs and SAEs after vaccination with AZD1222, it is proposed that any serious or severe events within the SOC of Nervous System disorders, including those of immunological origin, are included for close follow up in the RMP via routine and additional pharmacovigilance activity.

Further, due to the potential auto-immune aetiology in two events, the applicant is requested to investigate whether there may be potential molecular mimicry between the viral vector and human (neurologic) tissue.

To this end, the applicant may perform a Basic Local Alignment Search Tool (BLAST) search.

There was one case of anaphylaxis 63 days after vaccination which is not related to AZD1222. Relevantly, subjects with a history of allergic reactions (angioedema, anaphylaxis or allergic disease or reactions possibly exacerbated by any component of AZD1222, MenACWY or paracetamol) were excluded from participation in the studies. Therefore it may be that more immunologic reactions will be observed if the vaccine is used in the general public. Anaphylaxis has been included as a safety concern in the Safety Specification, and a warning is included in 4.4 to alert health care providers to this potential risk.

SAEs: Overall, the incidence of SAEs was low (less than 1%) and similar in the AZD1222 and control groups.

The most frequently reported SAEs by SOC in the AZD1222 and groups were Infections and Infestations (0.1% and 0.2% of participants respectively) and Injury, Poisoning and Procedure Related Complications (0.1% in both groups). There were 2 SAEs considered treatment related by the investigator in the AZD1222 group and 2 in the control (pyrexia, autoimmune haemolytic anaemia and myelitis).

There were 6 deaths (2 in the AZD1222 and 4 in the control groups), none related to the study intervention.

Safety by subgroup

The solicited and unsolicited AEs were less frequently reported in adults aged ≥65 years receiving any AZD1222 vaccine than in adults aged 18-64 years. Considering that the population aged ≥65 included for the assessment is much smaller than the group of adults aged 18-64, more data from older participants are needed to make any comparison on reactogenicity in the different age groups.

Regarding participants with comorbidity at baseline, a safety pool analysis has been provided showing a similar safety profile of AZD1222 in participants with and without comorbidities at baseline. Moreover, a subgroup analysis of safety stratified by different comorbidities (BMI ≥30kg/m2, cardiovascular disorder, respiratory disease, or diabetes) showed overall no imbalances in the incidence of solicited AEs, unsolicited AEs, SAEs and AESIs for any of the comorbidities.

Overall, slightly lower frequencies of solicited AEs were reported in seropositive than seronegative

participants at baseline, for both AZD1222 and control groups. However, the severity of the local AEs was observed in higher percentages in the seropositive population receiving AZD1222. Considering that the number of participants of seropositive exposed to AZD1222 is very low, these results should be interpreted with caution.

When analysing the reactogenicity profile of AZD1222 across countries, less frequent local and systemic AEs were reported in South Africa than in UK or Brazil, in both AZD1222 and control groups. There were slight differences in the percentage of severe solicited local and systemic AEs reported after receiving the AZD1222 in the South African population than in the UK or Brazil studies. Unsolicited AEs were most frequently

reported in Brazil as 98% of participants may have reported typical reactogenicity AEs as unsolicited AEs.

In general, no imbalances were observed between special populations, such as: age, comorbidity, country or serostatus regarding SAEs or AESIs.

There is only very limited clinical experience in pregnant women, with 14 pregnant women in the safety database who were exposed to AZD1222. Use of AZD1222 in pregnant and breastfeeding women will be investigated in the planned PASS activities.

Additional safety data needed in the context of a conditional MA

The final clinical study report for studies COV001, COV002, COV003 and COV005 will be submitted no later than May 2022. The primary analysis (based on the 7th December data cut-off (post data-base lock) and final analysis from the pooled pivotal studies will be submitted no later than March 2021 and May 2022 respectively.

In order to confirm the safety of AZD1222 in the elderly and subjects with underlying disease, the overview and summaries of the primary analysis and final clinical study report for study D8110C00001 will be

submitted no later than April 2021 and March 2024 respectively. These datasets are subject to specific obligations laid down in the MA.