Mechanisms of DNA damage across ADCs, radioligand therapies, and DDR inhibitors — and the case for genome-wide break mapping

Antibody–drug conjugates (ADCs), radioligand therapies (RLTs), and DNA damage response inhibitors (DDRi) are three of the most commercially and clinically active modalities in oncology. They are built on different engineering logic: a targeting antibody, a chelated radionuclide, a small-molecule kinase or enzyme inhibitor. Yet all three converge on the same terminal biochemical event — broken DNA — and all three lack an answer to the same unresolved question: not simply how much DNA damage a drug causes, but where in the genome that damage lands.^1,2,3

The field’s current technology toolkit answers the first question well and the second barely at all. γH2AX foci, comet assays, and mutational scar signatures are aggregate measurements that quantify burden.⁴ They do not report coordinates. As the modalities above move into combination regimens — ADC with DDRi payloads, RLT in conjunction with PARP inhibitors — that gap can quickly become the rate-limiting factor in rational trial design.

1. ADCs: precise delivery, imprecise lesions

Third-generation ADC development shifted the payload class from tubulin disruption to direct genotoxicity, largely because DNA-damaging payloads kill quiescent as well as cycling cells.¹ Three chemistries dominate and each produces a mechanistically distinct lesion.

Topoisomerase I inhibitor payloads — deruxtecan (DXd) in trastuzumab deruxtecan and datopotamab deruxtecan, SN-38 in sacituzumab govitecan — stabilize the Top1–DNA cleavage complex.⁵ The initial single-strand nick is not itself lethal, but can convert to a double-strand break when a replication fork or transcribing polymerase collides with the trapped complex, generating a one-ended DSB that requires homologous recombination to resolve.⁶ The break is therefore a product of collision geometry, not of the drug’s intrinsic chemistry, and tends to concentrate wherever replication and transcription are most active.

Calicheamicin, the enediyne payload in gemtuzumab and inotuzumab ozogamicin, behaves differently. It binds the DNA minor groove and causes oxidative-mediated backbone cleavage of duplex DNA in a site-specific manner, causing damage without requiring replication.⁷

Pyrrolobenzodiazepine (PBD) dimers form covalent guanine adducts and interstrand crosslinks.⁸ The DSB is not the primary lesion but a repair intermediate, produced when nucleotide excision and translesion synthesis machinery unhook the crosslink. The DNA break in this case is determined by the repair pathway, not by the payload.

Through a targeted approach, the antibody determines which cell receives the above payloads. The bystander effect — desirable for killing antigen-negative tumor cells — extends that efficacy to neighboring cells as well. However, nothing in an ADC’s design incorporates data on the distribution patterns of double-strand breakage that occurs once the payload is delivered to the target cells. What this added information can tell us may help identify toxic side effects as well as relevant PD and predictive biomarkers.

Where the biomarker field currently sits

Predictive biomarkers for Top1-payload ADCs remain surrogates rather than measurements. SLFN11 expression, BRCAness, and RB1 loss predict topoisomerase I inhibitor response in triple-negative breast cancer,⁹ and the ICARUS-LUNG01 phase 2 biomarker analysis of datopotamab deruxtecan found DNA repair pathway activation enriched in non-responders while SLFN11-positive tumor cells were more frequent among responders.¹⁰ These are informative signals indicative of the pathways involved in the response and repair of damage conferred by these payloads, which strongly contributes to treatment efficacy. But they primarily describe the cell’s capacity to respond to damage; none of them describes the pattern of damage itself.

A precedent that location matters

Topoisomerase II poisons make the case most forcefully. Etoposide and doxorubicin induce cleavage at defined consensus sites within the KMT2A (MLL) breakpoint cluster region and at the PML and RARA translocation breakpoints — the same junctions recovered from patients who later develop therapy-related acute myeloid and promyelocytic leukemia.^11,12 Sites of genomic alterations caused by these poisons are nonrandom, and it follows that the long-term liability of these agents is not just a function of how many breaks they make, but also a function of which sequences they cut.

2. Radioligand therapies: dose is anatomical, damage is genomic

RLT radiobiology is dominated by linear energy transfer (LET). Lutetium-177, a β⁻ emitter with LET near 0.2 keV/µm and a tissue path of up to ~2 mm, deposits energy sparsely; the majority of its DNA damage is indirect, mediated by water radiolysis and reactive oxygen species, and is dominated by single-strand breaks and base lesions with comparatively few DSBs. Actinium-225, an α emitter with LET in the 80–100 keV/µm range and a path length of 50–80 µm, deposits energy densely along a short track, producing clustered, complex double-strand breaks accompanied by adjacent base damage — lesions that are poorly handled by canonical repair.¹³

This introduces a large quantitative gap. Comparative γH2AX and viability work in somatostatin receptor–expressing models established a substantially higher relative biological effectiveness for ²²⁵Ac-DOTATOC versus ¹⁷⁷Lu-DOTATOC,¹⁴ and Geant4-DNA Monte Carlo modeling has estimated that ²²⁵Ac yields nearly two orders of magnitude more DSBs per gigabase per decay than ¹⁷⁷Lu, with yields rising sharply as the source approaches the nucleus.¹⁵ Clinically, α-emitter PSMA therapy has produced responses in patients refractory to β-emitter treatment, attributed to greater cell-specific effects and lower off-target toxicity.¹⁶

What RLT dosimetry does not provide are the genomic coordinates of recurring DSBs caused by the radioactive decay of these isotopes in the targeted cells. Absorbed dose is reported in Gy to a lesion or organ. Two patients with identical tumor-absorbed dose may sustain very different break landscapes, depending on chromatin accessibility, genomic aberrations, and repair genotype — however, standard dosimetry does not capture that.

Combination design is built on this asymmetry

The rationale for combining RLT with PARP inhibition follows from the logic that ¹⁷⁷Lu-PSMA-617 generates predominantly single-strand breaks that is repaired through PARP-dependent base excision repair, causing PARP blockade to convert those lesions into cancer-killing DSBs. The LuPARP phase 1 program in mCRPC established feasibility and tolerable scheduling, with radiographic response rates numerically higher than ¹⁷⁷Lu-PSMA-617 monotherapy; a parallel phase I established feasibility of ¹⁷⁷Lu-DOTATATE with olaparib in somatostatin receptor–positive tumors.^17,18 These studies are instructive: concurrent DDR blockade at the time of radioligand administration appears critical, and predictive biomarkers are more promising as a status of genomic instability and tumor burden than any observed genetic alteration.^17,19

3. DDR inhibitors: damage by subtraction

DDR inhibitors are mechanistically distinct from the first two classes. They do not deposit a double-strand break lesion. They remove the machinery that would otherwise resolve endogenous stress and let the genome accumulate breaks where loci and sequences are most vulnerable.

PARP inhibitors act principally through trapping — locking PARP1 onto DNA rather than merely inhibiting catalysis.²⁰ Trapped complexes at unligated Okazaki fragments generate single-stranded DNA gaps behind the replication fork, and gap accumulation is now regarded as a key determinant of synthetic lethality in BRCA-deficient cells.²¹ Fork stalling triggers reversal, and reversed forks are substrates for nucleolytic processing by MRE11, MUS81, EXO1, and SLX4 — leading to double-strand break formation.

ATR, CHK1, and WEE1 inhibitors operate one level up. Loss of ATR signaling permits unscheduled origin firing, depletes the nuclear supply of RPA, and drives replication catastrophe in which ssDNA at forks convert to double-stranded breaks en masse.²² WEE1 inhibition additionally leads to CDK1/2 hyperactivation, origin over-firing, and premature mitotic entry with unrepaired DNA damage.²³

Because the drug does not seed the lesion, the resulting break landscape is dictated by intrinsic genome architecture: replication timing, common fragile sites, R-loop-prone loci, non-B DNA structures, and repeat-rich late-replicating regions. Most of these features lie outside coding exons entirely.

The biomarkers used to select patients do not fully reflect this landscape of fragility exposed upon therapeutic targeting. BRCA1/2 status and HRD scores are scar-based — they read the mutational residue of historical repair deficiency.²⁴ They are retrospective by construction and reveal little about the acute dynamics of where the drug actively causes DNA breaks in the cancer cell at any given dosing, timepoint, and genome-wide landscape.

4. What current research is still missing

No genomic coordinates for therapeutic lesions

Measurement of γH2AX remains the field’s conventional pharmacodynamic readout, and its limitations are well documented: foci do not map one-to-one to DSBs, phosphorylation can also occur at single-stranded DNA breaks, chromatin conformation biases affect number and size, apoptotic H2AX phosphorylation confounds interpretation, and foci persist after breaks have been rejoined.^4,25 No assay in routine clinical or preclinical use reports the precise genomic coordinates of drug-induced breaks in a treated sample.

The non-coding blind spot

Nearly all validated DNA biomarkers exist as mutations in expressed genes — roughly 2% of the genome. The structural determinants of fragility that govern where a break occurs from acute therapeutic targeting are overwhelmingly non-coding. A biomarker framework restricted to the exome is therefore intrinsically unable to capture them.

DNA damage landscape variations across modalities

We cannot presently know whether different drug modalities and payloads result in distinctive DNA damage landscapes. Does ¹⁷⁷Lu break the same loci as ²²⁵Ac? Does a DXd payload versus an ATR inhibitor added to an ADC expand the break landscape into new regions — utilizing mechanistic complementarity — or simply intensify breakage at loci already being cut? The distinction has direct toxicity as well as efficacy implications and can vastly enhance combination and scheduling design.

Genotoxic liability is inferred, not measured

Therapy-related leukemia risk is currently established epidemiologically, years after exposure. Break-site data from hematopoietic or normal tissue would allow prospective assessment of whether a candidate agent cleaves at translocation-prone loci — a question the KMT2A precedent shows is answerable in principle.¹¹

The cytosolic compartment as an immunostimulant

Micronuclei and dispersed cytosolic dsDNA are the presumed connection between genotoxic therapy and cGAS–STING-driven immunogenicity,^26,27 but the picture has become contested: recent work reports that micronuclei formed from damage to the genome induced by radiation, replication stress, and segregation errors carry nucleosome-bound self-DNA that fail to activate cGAS–STING robustly.²⁸ Yet what DNA fragments reaches the cytosol, and from which genomic regions, is an open question with immediate bearing on RLT, DDR, and ADC combinations with immunotherapy.

5. What genome-wide break mapping adds

Methods for genome-wide DSB localization have matured considerably — from BLESS and BLISS to END-seq, DSBCapture, and DSB-Seq — and have already demonstrated that drug-induced breakage is strikingly non-random, clustering at transcription start sites, active promoters, and topoisomerase consensus sequences.^29-34 Translating that capability from mechanistic cell biology into drug development can advance the field of precision medicine.

This is the opportunity BreakSight® seeks to innovate in. Its main platform, DDsite, tags, retrieves, and sequences double-strand break sites genome-wide from cells exposed to a DNA damage–inducing treatment or condition. DDinsight processes that data through an analysis pipeline that maps and identifies reproducible break sites across the genome specific to a given treatment condition and compares break landscapes across conditions. DDinsight+ extends the analysis to repeats and sequence motifs enriched at break sites as well as towards any significant overlap of break sites with publicly available genomic datasets in the relevant cell type. DDsite-Cyto captures cytoplasmic breaks separately from nuclear ones. Applied to the modalities above, several concrete uses follow:

• Modality fingerprinting – A break-site map for a DXd payload, a calicheamicin payload, a β-emitter, an α-emitter, and a PARP inhibitor constitutes a mechanistic signature for each. It separates mechanism-based damage from off-target damage in a way that lesion counts cannot.
• Dose and schedule optimization – Break landscapes can be tested for saturation and redistribution across dose levels and administration sequences, which is the question LuPARP resolved empirically against clinical endpoints.
• Combination logic – Whether a DDRi added to an ADC or RLT expands the break-site map into new loci or amplifies existing ones is directly measurable, and it is the question that can discriminate a rational combination from an additive-toxicity one.
• Resistance monitoring – Longitudinal profiling can reveal whether resistance manifests as loss of breakage at previously conserved sites, redistribution to new regions, or unchanged breakage with restored repair — three mechanisms with three different clinical responses.
• Non-coding biomarker discovery – Recurrent break loci, enriched motifs, and DNA repeat classes become candidate stratification markers independent of the coding mutation catalogue — extending biomarker discovery into the 98% of the genome that expression-based approaches leave untouched.
• Immunogenicity and safety – Cytoplasmic break profiling provides a mechanistic input for immuno-oncology combination design, and break-site maps in normal or hematopoietic cells offer a prospective route to genotoxicity assessment.

6. A concrete next step

The most informative near-term experiment is a matched panel: one topoisomerase I–payload ADC, one β-emitter RLT, one α-emitter RLT, one PARP inhibitor, and one ATR or WEE1 inhibitor, profiled across a common isogenic model set spanning HR-proficient, BRCA-mutant, and SLFN11-null backgrounds, with paired nuclear and cytoplasmic break maps at multiple doses and time points.

That single dataset would establish whether break-site distribution is modality-specific, dose-dependent, and repair-genotype-dependent — three areas that currently gate rational combination design across all three classes. The field has spent two decades becoming better at counting DNA breaks. The next advance comes from knowing where they land.

References

1.  Fu Z, Li S, Han S, Shi C, Zhang Y. Antibody drug conjugate: the “biological missile” for targeted cancer therapy. Signal Transduct Target Ther. 2022;7(1):93. doi:10.1038/s41392-022-00947-7

2.  Sartor O, de Bono J, Chi KN, et al. Lutetium-177-PSMA-617 for Metastatic Castration-Resistant Prostate Cancer. N Engl J Med. 2021;385(12):1091-1103. doi:10.1056/NEJMoa2107322

3.  Groelly FJ, Fawkes M, Dagg RA, Blackford AN, Tarsounas M. Targeting DNA damage response pathways in cancer. Nat Rev Cancer. 2023;23(2):78-94. doi:10.1038/s41568-022-00535-5

4.  Löbrich M, Shibata A, Beucher A, et al. gammaH2AX foci analysis for monitoring DNA double-strand break repair: strengths, limitations and optimization. Cell Cycle. 2010;9(4):662-669. doi:10.4161/cc.9.4.10764

5.  Ogitani Y, Aida T, Hagihara K, et al. DS-8201a, A Novel HER2-Targeting ADC with a Novel DNA Topoisomerase I Inhibitor, Demonstrates a Promising Antitumor Efficacy with Differentiation from T-DM1. Clin Cancer Res. 2016;22(20):5097-5108. doi:10.1158/1078-0432.CCR-15-2822

6.  Pommier Y, Nussenzweig A, Takeda S, Austin CA. Human topoisomerases and their roles in genome stability and organization. Nat Rev Mol Cell Biol. 2022;23(6):407-427. doi:10.1038/s41580-022-00452-3

7.  Zein N, Sinha AM, McGahren WJ, Ellestad GA. Calicheamicin gamma 1I: an antitumor antibiotic that cleaves double-stranded DNA site specifically. Science. 1988;240(4856):1198-1201. doi:10.1126/science.3240341

8.  Hartley JA. Antibody-drug conjugates (ADCs) delivering pyrrolobenzodiazepine (PBD) dimers for cancer therapy. Expert Opin Biol Ther. 2021;21(7):931-943. doi:10.1080/14712598.2020.1776255

9.  Coussy F, El-Botty R, Château-Joubert S, et al. BRCAness, SLFN11, and RB1 loss predict response to topoisomerase I inhibitors in triple-negative breast cancers. Sci Transl Med. 2020;12(531):eaax2625. doi:10.1126/scitranslmed.aax2625

10.  Planchard D, Cozic N, Mosele MF, et al. Efficacy, safety, and biomarker analysis of datopotamab deruxtecan in advanced non-small cell lung cancer: ICARUS-LUNG01 phase 2 study. Cancer Cell. 2026;44(6):1147-1159.e6. doi:10.1016/j.ccell.2026.03.017

11.  Libura J, Slater DJ, Felix CA, Richardson C. Therapy-related acute myeloid leukemia-like MLL rearrangements are induced by etoposide in primary human CD34+ cells and remain stable after clonal expansion. Blood. 2005;105(5):2124-2131. doi:10.1182/blood-2004-07-2683

12.  Mistry AR, Felix CA, Whitmarsh RJ, et al. DNA topoisomerase II in therapy-related acute promyelocytic leukemia. N Engl J Med. 2005;352(15):1529-1538. doi:10.1056/NEJMoa042715

13.  Sgouros G, Roeske JC, McDevitt MR, et al. MIRD Pamphlet No. 22 (abridged): radiobiology and dosimetry of alpha-particle emitters for targeted radionuclide therapy. J Nucl Med. 2010;51(2):311-328. doi:10.2967/jnumed.108.058651

14.  Graf F, Fahrer J, Maus S, et al. DNA double strand breaks as predictor of efficacy of the alpha-particle emitter Ac-225 and the electron emitter Lu-177 for somatostatin receptor targeted radiotherapy. PLoS One. 2014;9(2):e88239. doi:10.1371/journal.pone.0088239

15.  Chatzipapas KP, Papachristou K, Visvikis D, Incerti S, Hazle JD, Kagadis GC. In silico assessment of cellular damage from Lu-177, Ac-225, and Pb-212 therapeutic radionuclides. Med Phys. 2025;52(11):e70089. doi:10.1002/mp.70089

16.  Kratochwil C, Bruchertseifer F, Giesel FL, et al. 225Ac-PSMA-617 for PSMA-Targeted α-Radiation Therapy of Metastatic Castration-Resistant Prostate Cancer. J Nucl Med. 2016;57(12):1941-1944. doi:10.2967/jnumed.116.178673

17.  Sandhu S, et al. LuPARP: phase 1 trial of 177Lu-PSMA-617 and olaparib in patients with metastatic castration-resistant prostate cancer. J Clin Oncol. 2023;41(16):5005-5005. doi:10.1200/JCO.2023.41.16_suppl.5005

18.  Hallqvist A, Brynjarsdóttir E, Krantz T, Sjögren M, Svensson J, Bernhardt P. ¹⁷⁷Lu-DOTATATE in Combination with PARP Inhibitor Olaparib Is Feasible in Patients with Somatostatin-Positive Tumors: Results from the LuPARP Phase I Trial. J Nucl Med. 2025;66(5):707-712. doi:10.2967/jnumed.124.268902

19.  Lückerath K, Mínguez Gabiña P, Eberlein U, et al. Radiopharmaceutical Therapy Meets Radiobiology: Implications for Trial Design and DNA Repair Inhibitor Combinations. J Nucl Med. 2026;67(8):1181-1188. doi:10.2967/jnumed.125.271527

20.  Murai J, Huang SY, Das BB, et al. Trapping of PARP1 and PARP2 by Clinical PARP Inhibitors. Cancer Res. 2012;72(21):5588-5599. doi:10.1158/0008-5472.CAN-12-2753

21.  Cong K, Peng M, Kousholt AN, et al. Replication gaps are a key determinant of PARP inhibitor synthetic lethality with BRCA deficiency. Mol Cell. 2021;81(15):3128-3144.e7. doi:10.1016/j.molcel.2021.06.011

22.  Toledo LI, Altmeyer M, Rask MB, et al. ATR prohibits replication catastrophe by preventing global exhaustion of RPA. Cell. 2013;155(5):1088-1103. doi:10.1016/j.cell.2013.10.043

23.  Elbæk CR, Petrosius V, Sørensen CS. WEE1 kinase limits CDK activities to safeguard DNA replication and mitotic entry. Mutat Res. 2020;819-820:111694. doi:10.1016/j.mrfmmm.2020.111694

24.  Telli ML, Timms KM, Reid J, et al. Homologous Recombination Deficiency (HRD) Score Predicts Response to Platinum-Containing Neoadjuvant Chemotherapy in Patients with Triple-Negative Breast Cancer. Clin Cancer Res. 2016;22(15):3764-3773. doi:10.1158/1078-0432.CCR-15-2477

25.  Rothkamm K, Barnard S, Moquet J, Ellender M, Rana Z, Burdak-Rothkamm S. DNA damage foci: Meaning and significance. Environ Mol Mutagen. 2015;56(6):491-504. doi:10.1002/em.21944

26.  Harding SM, Benci JL, Irianto J, Discher DE, Minn AJ, Greenberg RA. Mitotic progression following DNA damage enables pattern recognition within micronuclei. Nature. 2017;548(7668):466-470. doi:10.1038/nature23470

27.  Mackenzie KJ, Carroll P, Martin CA, et al. cGAS surveillance of micronuclei links genome instability to innate immunity. Nature. 2017;548(7668):461-465. doi:10.1038/nature23449

28.  Takaki T, Millar R, Hiley CT, Boulton SJ. Micronuclei induced by radiation, replication stress, or chromosome segregation errors do not activate cGAS-STING. Mol Cell. 2024;84(11):2203-2213.e5. doi:10.1016/j.molcel.2024.04.017

29.  Crosetto N, Mitra A, Silva MJ, et al. Nucleotide-resolution DNA double-strand break mapping by next-generation sequencing. Nat Methods. 2013;10(4):361-365. doi:10.1038/nmeth.2408

30.  Yan WX, Mirzazadeh R, Garnerone S, et al. BLISS is a versatile and quantitative method for genome-wide profiling of DNA double-strand breaks. Nat Commun. 2017;8:15058. doi:10.1038/ncomms15058

31.  Canela A, Sridharan S, Sciascia N, et al. DNA Breaks and End Resection Measured Genome-wide by End Sequencing. Mol Cell. 2016;63(5):898-911. doi:10.1016/j.molcel.2016.06.034

32.  Baranello L, Kouzine F, Wojtowicz D, et al. DNA break mapping reveals topoisomerase II activity genome-wide. Int J Mol Sci. 2014;15(7):13111-13122. doi:10.3390/ijms150713111

33.  Mourad R, Ginalski K, Legube G, Cuvier O. Predicting double-strand DNA breaks using epigenome marks or DNA at kilobase resolution. Genome Biol. 2018;19(1):34. doi:10.1186/s13059-018-1411-7

34.  Saayman X, Esashi F. Breaking the paradigm: early insights from mammalian DNA breakomes. FEBS J. 2022;289(9):2409-2428. doi:10.1111/febs.15849