Antibody-DrugAntibody-DrugConjugates: SimpleConjugates: SimpleIdea, ComplicatedIdea, Complicated

MatterMatterHaleh SaberHaleh Saber

John K. LeightonJohn K. Leighton

Of�ce of Oncology Disease, Division of Hematology OncologyOf�ce of Oncology Disease, Division of Hematology OncologyToxicologyToxicology

US FDAUS FDA

A ntibody-drug conjugates (ADCs) are a class ofpharmaceuticals that consist of small moleculedrugs (also known as payloads) covalently attached

to an antibody via a linker (1,2). While the majority of theseproducts are in oncology, occasionally they are used in non-

oncology indications, such as in rheumatoid arthritis (3) andinfectious diseases (4). This review will discuss the use of thecurrent generation of ADCs in oncology (i.e., ADCs thatcontain cytotoxic payloads) and excludes any discussion onconjugated products containing toxins, such asmoxetumomab pasudotox (5). Additional information isavailable in ICH S9 Guidance and ICH S9 Questions andAnswers for nonclinical development of ADCs in oncology(6,7) and in ICH S6 Guidance for nonclinical development ofbiotechnology-derived products (8).

Targeting cancer cells with anti-cancer agents while sparing healthy tissues is an attractiveidea, though achieving this goal is complicated. Challenges include but are not limited to on-target/off-tumor binding, release of the payload outside of tumor cells, and release ofdeconjugated payload from tumor cells, resulting in off-site toxicities. Progress has beenmade over the last few years with increased understanding of tumor biology and improvedtechnology. Newer versions of ADCs are emerging using new targets, optimizing the linkerand drug-antibody ratios, and incorporating site-speci�c payload attachment to engineeredantibodies (1,9,10).

While the ADC technology has evolved, many ADCs are designed using the same linker-payload, such as vcMMAE, mcMMAF, SMCC-DM1, and vaPBD (11,12,13). This may be partiallydue to the availability of technology that makes it easier to generate a new ADC and thepotential concern that any new technology may be associated with uncertainties andundesired outcomes. In addition, using the same data for multiple regulatory submissionsmakes it less costly and faster to submit a new IND (Investigational New Drug application) ora marketing application by eliminating conduct of certain nonclinical studies needed insupport of regulatory submissions.

Challenges in ADC Development

As of September 2020, there were nine ADCs approved by the US Food and DrugAdministration (FDA; see table below).

ADCPayload

(Mechanism)Indication

Year

ApprovedReferences

Adcetris

(brentuximab

vedotin)

MMAE

(microtubule

disrupting)

Hodgkin

lymphoma

(HL)

Anaplastic

large cell

lymphoma

(ALCL)

2011 Seattle Genetics,

2020

Peripheral T-

cell

lymphoma

(PTCL)

Kadcyla (ado-

trastuzumab

emtansine)

DM1

(microtubule

disrupting)

Breast cancer 2013Genentech,

2020a

Mylotarg

(gemtuzumab

ozogamicin)

Calicheamicin

(DNA

damaging)

Acute

myeloid

leukemia

(AML)

2017*Wyeth Pharms

Inc, 2020a

Besponsa

(inotuzumab

ozogamicin)

Calicheamicin

Acute

lymphoblastic

leukemia

(ALL)

2017Wyeth Pharms

Inc, 2020b

Polivy

(polatuzumab

vedotin-piiq)

MMAE

Diffuse large

B-cell

lymphoma

(DLBCL)

2019Genentech,

2020b

Padcev

(enfortumab

vedotin-ejfv)

MMAEUrothelial

cancer2019 Astellas 2020

Enhertu (fam-

trastuzumab

deruxtecan-

nxki)

DXd

(topoisomerase

inhibitor)

Breast cancer 2019Daiichi Sankyo,

2020

Trodelvy

(sacituzumab

govitecan-

hziy)

SN-38

(topoisomerase

inhibitor)

Breast cancer 2020Immunomedics

Inc, 2020

Blenrep

(belantamab

mafodotin-

blmf)

MMAF

(microtubule

disrupting)

Multiple

myeloma2020

GlaxoSmithKline,

2020

ADC Adcetris (brentuximab vedotin)

Payload (Mechanism) MMAE (microtubule disrupting)

Indication

Hodgkin lymphoma (HL)

Anaplastic large cell lymphoma (ALCL)

Peripheral T-cell lymphoma (PTCL)

Year Approved 2011

References Seattle Genetics, 2020

ADC Kadcyla (ado-trastuzumab emtansine)

Payload (Mechanism) DM1 (microtubule disrupting)

Indication Breast cancer

Year Approved 2013

References Genentech, 2020a

ADC Mylotarg (gemtuzumab ozogamicin)

Payload (Mechanism) Calicheamicin (DNA damaging)

Indication Acute myeloid leukemia (AML)

Year Approved 2017*

References Wyeth Pharms Inc, 2020a

ADC Besponsa (inotuzumab ozogamicin)

Payload (Mechanism) Calicheamicin

Indication Acute lymphoblastic leukemia (ALL)

Year Approved 2017

References Wyeth Pharms Inc, 2020b

ADC Polivy (polatuzumab vedotin-piiq)

Payload (Mechanism) MMAE

Indication Diffuse large B-cell lymphoma (DLBCL)

Year Approved 2019

References Genentech, 2020b

ADC Padcev (enfortumab vedotin-ejfv)

Payload (Mechanism) MMAE

Indication Urothelial cancer

Year Approved 2019

References Astellas 2020

ADC Enhertu (fam-trastuzumab deruxtecan-nxki)

Payload (Mechanism) DXd (topoisomerase inhibitor)

Indication Breast cancer

Year Approved 2019

References Daiichi Sankyo, 2020

ADC Trodelvy (sacituzumab govitecan-hziy)

Payload (Mechanism) SN-38 (topoisomerase inhibitor)

Indication Breast cancer

Year Approved 2020

References Immunomedics Inc, 2020

ADC Blenrep (belantamab mafodotin-blmf)

Payload (Mechanism) MMAF (microtubule disrupting)

Indication Multiple myeloma

Year Approved 2020

References GlaxoSmithKline, 2020

* First approved in 2000, but the application was subsequently withdrawn in 2010. The product was

then re-approved in 2017.

While the initial emphasis was mostly in hematologic malignancies, interest has spiked insolid tumors with three recent FDA approvals in breast and urothelial cancers. While cells ofthe hematopoietic system are readily accessible, solid tumors may require additional steps ofextravasation of the ADC and penetration into the solid tumor by overcoming the stromalbarrier (14). To our knowledge, no information is currently available in the scienti�c literatureproviding a side-by-side, data-driven comparison of the challenges associated with ADCdevelopment in hematologic malignancies and solid tumors. Challenges related to thedelivery of ADCs to tumor cells might have contributed to an initial slow development ofADCs in solid tumors. More recent articles point to a growing number of ADCs indevelopment for solid tumors (15,16,17).

A desired outcome is to deliver the ADCs to tumor cells and spare healthy tissues of toxicitiesassociated with the payloads, but the payloads will eventually be released and re-distribute.The product label of FDA-approved ADCs and additional data analysis for the currentgeneration of ADCs (11,12,18) indicate that dose-limiting toxicities (DLTs) in animals andpatients are mainly related to the payload. These include but are not limited to toxicities atdistant sites, such as hepatotoxicity observed with Kadcyla, Adcetris, and Polivy, and oculartoxicities associated with Blenrep. The maximum tolerated doses (MTDs) in patients havebeen dependent on the payload and can be generally predicted for ADCs using the samelinker-payload, drug-to-antibody ratio (DAR), and frequency of administration. The humanMTDs were in the range of 1.8 to 2.4 mg/kg with vcMMAE-conjugated ADCs that had a DAR of4 and were given every three weeks as an intravenous infusion (11). When the payload isassociated with immune activation and pro-in�ammatory responses, high inter-subject

variability may be encountered, leading to a wider range of human MTDs for comparableADCs and dosing regimen.

In 2019, we reported the dif�culties pharmaceutical companies were facing whendetermining the human MTDs in clinical trials of pyrrolobenzodiazepine (PBD)-ADCs when 15separate INDs were examined (12). Our analysis indicated that PBD caused multi-organin�ammatory responses. Immune-mediated events can have delayed onset and affectvarious organs. In addition, they may be prone to inter-subject variability and may varydepending on the genetics, previous antigen exposures, and previous anticancer treatments.The many factors contributing to the onset, nature, and severity of immune-mediated�ndings can make it dif�cult to select a dose that will be tolerated across indications andpatient populations, even for ADCs with the same payload and DAR, and when using thesame schedule of administration.

Another example of an immune-activating payload may be the auristatin MMAF. The datasetevaluated by the FDA in 2015 (11) included only two MMAF-ADCs; thus a thorough evaluationof toxicities was not possible. With more data becoming available, signals of pro-in�ammatory responses are being detected with MMAF-ADCs and attributed to MMAF.Blenrep (belantamab mafodotin) is an MMAF-containing ADC recently approved for thetreatment of patients with multiple myeloma. Based on toxicology data (FDA multi-

disciplinary review for Blenrep, 2020), treatment of animals with belantamab mafodotin orthe unconjugated payload resulted in pro-in�ammatory responses as indicated by changesin hematology parameters and histopathology observations of multi-organ in�ammation.

One of the notable �ndings with Blenrep described in the product label is ocular toxicity(belantamab mafodotin). Ocular toxicity was also noted in animals treated with both thepayload or belantamab mafodotin, and while its cause is not entirely understood, wespeculate that MMAF-related in�ammatory responses may have contributed to the �ndingsin animals. Although ocular toxicities have been previously described in patients treated withADCs (19), these �ndings appear to be more evident and generally of higher severity inMMAF-ADCs and maytansinoid DM4-conjugated ADCs, with the following terms used indescribing the events in patients: blurred vision, dry eye, keratopathy, microcystickeratopathy, keratitis, iridocyclitis, corneal epitheliopathy, eye pain, conjunctival hemorrhage,corneal deposits, and photophobia. It has been suggested that the difference between oculartoxicities associated with MMAF-ADCs and MMAE-ADCs may be related to the charged (formcMMAF) versus uncharged (for vcMMAE) metabolites of auristatins and thus differences intheir cell permeability (18,19,20,21). But the discussions do not explain the mechanism ofocular �ndings or why the eyes are more sensitive than other organs. Additional studies willbe needed to better understand the cause of the ocular toxicities.

In a cross-biologic license application (BLA) comparison of FDA-approved MMAF-ADCs andMMAE-ADCs, we noted more pronounced ocular �ndings in animals treated withbelantamab mafodotin compared to the three approved MMAE-ADCs brentuximab vedotin,polatuzumab vedotin, and enfortumab vedotin (FDA Pharmacology Review for Adcetris,

2011; FDA Pharmacology Review for Polivy, 2019; FDA multi-disciplinary review for Padcev,

2019; FDA multi-disciplinary review for Blenrep, 2020). The same pattern is noted in patients,based on the information available in the product labels.

Circulating ADCs with non-speci�c distribution may result in early onset of toxicities. This isbased on the observation that payload-related toxicities were more evident in rodent studies

where no binding to the target occurred; toxicities were less prominent or delayed inmonkeys where the ADCs bound to their targets (11). This was also seen in the review ofnonclinical data for belantamab mafodotin (FDA multi-disciplinary review for Blenrep) thatnoted more pronounced ocular toxicities of the ADC in rodents (no target binding) comparedto the studies of the ADC in monkeys (binding occurs). These observations suggest that highlevels of target expression may delay the emergence of payload-related toxicities.

Conclusion

In summary, there is a growing number of ADCs in development for both solid tumors andhematologic malignancies, and there have been several recent regulatory approvals. WhileADCs are an attractive platform for therapeutic intervention, their full potential may not yetbe realized. Despite advances made over the years, the current generation of ADCs continueto demonstrate DLTs related to the payload, resulting in advantages and disadvantages inproduct development.

Accumulated clinical and non-clinical safety data from speci�c linker-payload platforms haveresulted in more ef�cient nonclinical product development by reducing the number ofanimal studies needed to characterize payload-related toxicities. This accumulatedknowledge can also lead to more ef�cient dose escalation trial designs by eliminatingsubtherapeutic dose levels of the ADCs when human MTDs have been reported to be in atight range regardless of the antibody, as was the case for vcMMAE-ADCs. This data can alsobe used to improve safety of dose escalation by urging caution in escalating beyond a setpoint based on experience with related products.

Payload-related toxicities continue to create challenges in the development of ADCs asdescribed above. Despite advances in product stability to reduce deconjugation of payloadsin the plasma and the selection of better targets (e.g., tumor antigens with minimalexpression in healthy tissues), the payloads are eventually released and redistributed, and attimes can result in severe toxicities that may halt or delay product development. Thechallenges associated with the current generation of ADCs have led to the birth of novel ideassuch as non-toxic payloads that could be converted to an active drug with additional triggers.One such design is described in a recent article for photo-activatable ADCs (22).

New platforms can be complicated, and FDA remains ready to engage with sponsors early ina product’s development to advance safe and effective treatments for patients with cancer.

References 1 to 22 available upon request.