Discussion on non-clinical aspects

2.5.4.7. Local Tolerance

Local tolerance was assessed in both single and repeat-dose toxicity studies. When changes were observed, these were considered procedurally related and similar in incidence and severity between control and tremelimumab dosed animals.

2.5.4.8. Other toxicity studies Tissue cross reactivity

Tissue cross reactivity studies of tissue binding of a fluorosceinated version of tremelimumab to cynomolgus monkey and human tissues was presented in reports IM645 and IM676. The studies were conducted according to GLP at Pathology Associates, a Charles River Company, Maryland, USA. The range of tissue was sufficiently broad and covered tissues of vital organs such organs of reproduction, heart and lung apart from expected target organs of gastrointestinal system, thymus, pancreas and lymph system. Human lymphocytes and human cerebellum tissue were used as positive and negative control, respectively.

The tissue binding profile of the two species was remarkably similar. The tissues binding tremelimumab were tonsils, lymphocytes in stomach, colon, spleen, lymph nodes and thymus in monkey. In human tissues it was tonsils, lymph nodes, thymus, lymphocytes in spleen, colon and small intestine with low binding in 1 out of three donors of thyroid. Tissue binding correlates with expected pharmacological effect and adverse findings in the monkey and adverse effects in patients.

Antigenicity

Tremelimumab did give rise to antidrug antibodies in the monkeys, however with limited impact on exposure. Only few animals showed decreasing exposure over time due to neutralising antidrug antibodies. This seems to be the case in patients as well, where 12.1% tested positive for ADAs and 10.0% for neutralising ADAs. The presence of ADAs did not impact tremelimumab pharmacokinetics, and there was no apparent effect on efficacy and safety (SmPC).

Immunotoxicity

Tremelimumab is a product, which enhance the reactivity of the immune system by inhibiting one of the down-regulating functions (CTLA4). This gives rise to general inflammation (in essence

autoimmune reactions) in a range of organs - most severely in the intestinal system and skin as observed from clinical signs in the monkey. The increase in general inflammation seems to be well documented in the studies in cynomolgus monkeys also on the cellular level but may be less obvious in the patient population in which leucopenia and neutropenia are very common adverse effects.

It is acknowledged that tremelimumab inhibits CTLA4 and thereby activate T cells. A range of both in vitro, in vivo pharmacology and repeat-dose toxicity studies documents this effect, which in vivo translates into severe systemic inflammation and mortality after repeat-dosing. However, the lack of effects on Tregs ability to dampen IFN-γ production by activated T cells is a concern. According to e.g.

Ohue, 2019, Tregs may be part of cancer tumours microenvironment to enhance tumour immunity providing a possibility for evading the activated T cells.

To further explain this trait of tremelimumab not targeting the intratumoral Tregs limiting its efficacy in cancer treatment, a scientific discussion was provided. Depletion of Tregs is dependent on ADCC of which tremelimumab is not capable mainly due to lack of FcR affinity. Selby et al. demonstrated that in mouse tumour models surrogate antibodies with higher affinity for FcR showed both the ability of depleting Tregs and enhanced antitumour activity.

There is a difference in affinity of IgG isotypes for FcR between mouse and human. IgG2a is a mouse isotype, with relatively potent Fc binding properties and is broadly equivalent to human IgG1.

Additionally, human IgG2 (such as tremelimumab) has very minimal Fc binding properties and is broadly equivalent to mouse IgG1 (as used in the in vivo studies described below) (Stewart et al 2014).

This discrepancy between nonclinical and clinical findings could be summarised as translational challenges associated with: 1) differences between IgG isotypes across species; 2) type of effector cells infiltrated in tumour and expression of different FcγRs on the surface between mouse and human;

3) varying CTLA-4 expression level on Tregs.

To conclude, tremelimumab is not capable of performing ADCC and therefore does not reduce Tregs number. In the context of immune related adverse events, that property is desirable, but intratumoral Tregs might be potential target for more efficient therapy because reducing Tregs inside tumours is associated with superior antitumour activity. Tremelimumab achieves its effect by targeting CTLA-4 on activated effector T cells and should be administered in combination with anti-PD-L1 antibody. Results from nonclinical studies showed that combination is superior to monotherapy with tremelimumab in cancer treatment, but similar to anti-PD-L1 monotherapy. Totality of data suggest that not affecting Tregs might be the reason for weaker efficacy of tremelimumab.

Key in vitro and in vivo studies highlight applicant´s statement that tremelimumab is not capable of affecting Tregs, however, the absence might be associated with weaker clinical outcomes and questionable contribution of tremelimumab in antitumour efficacy.

This deficiency might also explain the modest effect in the in vivo mouse cancer models.

Despite the principle of abrogating the T-cell inhibition was studied in vitro as well as in vivo, different results have been obtained from different animal disease models. A proof of principle specific for the tumour type included in the indication was not demonstrated. Thus, the relevance for the current proposed indication, uHCC, was not provided. According to e.g. Chulpanova et al 2020, the syngeneic mouse tumour models lack the complexity of the tumour microenvironment observed in patients, hence the translational value of the mouse model is questionable.

Pharmacokinetics

As expected for a monoclonal antibody, volume of distribution is mostly confined to the vascular space as the volume of distribution in monkey demonstrate (Vss = 54 mL/kg). The major elimination

pathway of tremelimumab is expected to be through protein catabolism. Pharmacokinetic drug-drug interactions of tremelimumab with other therapeutics are not anticipated.

The pharmacokinetics of tremelimumab appear to be independent of dose and providing a linear relation between exposure and dose within the dose range tested (0.75 to 100 mg/kg), which also is the case in humans over the dose range of 1 to 20 mg/kg.

Antidrug antibodies (ADAs) were observed in several animals during the repeat-dose toxicology studies and in some cases appeared to increase clearance. However, the overall exposure was deemed

sufficient securing the validity of the studies.

Toxicology

Repeat-dose toxicity studies were conducted in monkeys of 1- or 6- months duration. In the 1-mont study findings were consistent with tremelimumab pharmacology by inducing inflammation but not severe.

In the chronic 6-month study in cynomolgus monkeys, treatment with tremelimumab was associated with dose-related incidence in persistent diarrhoea and skin rash, scabs and open sores, which were dose-limiting. These clinical signs were also associated with decreased appetite and body weight and swollen peripheral lymph nodes. Histopathological findings correlating with the observed clinical signs included reversible chronic inflammation in the cecum and colon, and mononuclear cell infiltration in the skin and hyperplasia in lymphoid tissues. A dose-dependent increase in the incidence and severity of mononuclear cell infiltration with or without mononuclear cell inflammation was observed in the salivary gland, pancreas (acinar), thyroid, parathyroid, adrenal, heart, oesophagus, tongue, periportal liver area, skeletal muscle, prostate, uterus, pituitary, eye (conjunctiva, extra ocular muscles), and choroid plexus of the brain. No NOAEL was found in this study with animals treated with the lowest dose of 5 mg/kg/week requiring supportive care. This dose provided an exposure-based safety margin of 1-2 to clinically relevant exposure (taking species difference in potency into account).

Mononuclear cell infiltration in prostate and uterus was observed in repeat dose toxicity studies. Since animal fertility studies have not been conducted with tremelimumab, the clinical relevance of these findings for fertility is unknown. In reproduction studies, administration of tremelimumab to pregnant Cynomolgus monkeys during the period of organogenesis was not associated with maternal toxicity or effects pregnancy losses, foetal weights, or external, visceral, skeletal abnormalities or weights of selected foetal organs.

Tremelimumab potential for influencing fertility and early embryonic development was not evaluated or discussed by the applicant. According to ICH S9, effects on reproductive organs from the repeat-dose toxicity studies can make the basis for this evaluation.

Pre- and postnatal development studies were not performed, and this is acceptable and in line with ICH S9.

No studies in juvenile animals were performed, and this is acceptable since the sought indication is only including adult patents.

Tremelimumab was not evaluated for genotoxic potential, and this is acceptable for a monoclonal antibody. Carcinogenic potential of tremelimumab was not evaluated, and this is acceptable given the indication sought in the treatment of unresectable hepatocellular carcinoma (uHCC).

RMP

The findings observed in the pivotal repeat-dose general toxicity studies of inflammation in cecum, colon and skin were also observed in patients. Moreover, clinical chemistry findings in patients and monkeys related to liver toxicity correlated to histological changes. As for toxicity to reproduction, it is acknowledged that the EFD study in monkeys did not give rise to concerns. However, inflammatory

markers were present in organs of reproduction of both male and female animals even after 99 days of recovery.