Future of radiotheranostics

Abstract

Radiotheranostics is reshaping precision oncology by integrating molecular imaging with targeted radionuclide therapy through matched diagnostic–therapeutic radiopharmaceutical pairs. Landmark clinical successes, including Lutetium-177-DOTATATE (¹⁷⁷Lu-DOTATATE) and ¹⁷⁷Lu-prostate-specific membrane antigen-617 (¹⁷⁷Lu-PSMA-617), have validated this approach and accelerated the shift from empiric dosing towards evidence-based practice supported by randomised trials. Beyond established somatostatin receptor and PSMA paradigms, emerging platforms targeting tumour–stroma and lineage-associated biology, such as fibroblast activation protein and other receptor systems, are broadening applications across diverse malignancies. In parallel, advances in radionuclide science, including next-generation β-emitters and targeted α-particle therapies, are enabling strategies with higher biological effectiveness and potential advantages in small-volume and micrometastatic disease. Innovations in ligand and vector engineering (including half-life extension and multivalent constructs) aim to improve tumour retention and therapeutic index. Finally, quantitative imaging, personalised dosimetry (from organ-based to voxel-level approaches) and artificial intelligence (AI)-enabled workflows are laying the foundation for scalable, individualised treatment planning and response assessment. This review synthesises key developments in molecular targeting, radiochemistry, dosimetry, AI and combination strategies, outlining the priorities required to translate next-generation radiotheranostics into durable, multidisciplinary cancer care.

Introduction

Radiotheranostics integrates diagnostic imaging with targeted radionuclide therapy (TRT), utilising molecularly matched radiotracers for precise tumour visualisation and treatment. This paradigm relies on ‘companion diagnostics’ where a tracer like Gallium-68–labelled prostate-specific membrane antigen ligand (⁶⁸Ga-PSMA-11) enables patient selection for Lutetium-177 (¹⁷⁷Lu)-PSMA-617 therapy, exemplifying theranostic synergy.¹ This targeted approach helps confine radioactive effects primarily to tumour sites while reducing overall systemic toxicity.²

The clinical success of theranostic pairs such as Lutathera/Netspot and Pluvicto/Locametz has generated substantial interest as treatment options for patients with refractory and metastatic cancers who have exhausted conventional therapies.^{1 3} Early development of radiotheranostics was largely guided by empirical dose-escalation and single-institution experiences, with limited multicentre validation.^{4 5} Over the past decade, however, the field has transitioned to evidence-based clinical adoption supported by randomised trials, most notably the phase III VISION study for ¹⁷⁷Lu-PSMA-617 and the Neuroendocrine Tumors Therapy (NETTER-1) trial for ¹⁷⁷Lu-DOTATATE, which established efficacy in refractory disease.^{6 7 8} Current modelling from the Lancet Oncology Commission on radiotherapy and theranostics estimates that more than 158 000 patients annually may be eligible for ¹⁷⁷Lu-PSMA therapy, requiring a global supply chain capable of delivering over 553 000 doses per year and underscoring the rapidly expanding infrastructural demands on nuclear medicine.⁹

In this review, we argue that radiotheranostics is undergoing a qualitative transition—from empirically delivered, isotope-centred treatments towards a modular, data-driven oncology platform that integrates molecular biology, quantitative imaging, dosimetry and digital decision support. We synthesise advances in target biology, radionuclide and ligand engineering, and personalised treatment planning, and place them in the context of evolving clinical trial paradigms, regulatory frameworks and health-system constraints. Rather than providing a catalogue of agents, we focus on the strategic directions that will determine whether radiotheranostics matures into a scalable component of precision oncology: biologically informed targeting, quantitative personalisation, rational combination strategies and system-level readiness for global implementation.

Expanding molecular targets beyond PSMA and SSTR and emerging tumour-associated targets

Fibroblast activation protein inhibitor

Fibroblast activation protein (FAP) is a serine protease highly expressed in the stroma of more than 90% of epithelial tumours, including lung, breast and pancreatic cancers, while remaining largely absent in healthy tissues.10,12 FAP inhibitors (FAPI), typically quinoline-based small molecules, target cancer-associated fibroblasts, key mediators of the tumour microenvironment, metastasis and immune evasion, can be radiolabelled for TRT.¹⁰

A landmark systematic review analysing 27 studies and 144 patients with advanced malignancies highlighted the clinical potential of lutetium-177-labeled fibroblast activation protein inhibitor(¹⁷⁷Lu-FAPI) therapy, reporting disease control rates ranging from 18.2% to 83.3% with favourable tolerability, as only about 10% of patients experienced grade 3–4 toxicities.¹³ However, most available data derive from heavily pre-treated populations, and the true efficacy of FAPI-based TRT in earlier disease stages remains to be defined.¹³

In soft tissue sarcomas, where FAP is expressed in over 90% of cases, early clinical experience suggests that ¹⁷⁷Lu-FAPI therapy can achieve disease stabilisation in a substantial proportion of patients with otherwise limited options, while maintaining a favourable safety profile.¹⁴

Most FAP-TRT studies have used β-emitters, which provide a crossfire effect to irradiate adjacent tumour cells. Although α-emitters such as ²²⁵Ac induce more potent DNA damage, their short range may result in incomplete tumour coverage; this has prompted interest in tandem α/β-emitter strategies to optimise efficacy.¹³ More recently, ²²⁵Ac-3BP-3940 has been explored in therapy-refractory metastatic colorectal cancer, either alone or in combination with β-emitters, demonstrating encouraging tolerability with mainly low-grade, transient toxicities.¹⁵

Chemokine receptor type 4

The chemokine receptor type 4 (CXCR4) and its ligand CXCL12 play central roles in hematopoietic trafficking and in tumour angiogenesis and metastasis, particularly through chemotactic recruitment of malignant cells to CXCL12-rich sites such as the bone marrow.^{16 17} Targeting CXCR4, therefore, offers a strategy to disrupt pathways critical for tumour dissemination. Pentixather has emerged as a promising ligand for TRT; however, because CXCR4-directed endoradiotherapy can induce profound myeloablation, its clinical use is currently largely restricted to settings where stem cell transplantation is feasible.¹⁸

First-in-human studies of ¹⁷⁷Lu/⁹⁰Y-pentixather in heavily pretreated multiple myeloma showed high tumour retention and radiation dose with meaningful metabolic responses, but predictable marrow toxicity requiring autologous stem cell support.¹⁹ Subsequent studies extended this concept to aggressive lymphomas and T-cell malignancies, confirming feasibility and potential efficacy in patients lacking alternative treatment options.^{20 21} More recently, retrospective data suggest that combining CXCR4-targeted endoradiotherapy with chemotherapy and total body irradiation may provide a personalised salvage strategy in multirefractory acute myeloid leukaemia without compromising stem cell engraftment.²²

Gastrin-releasing peptide receptor

High gastrin-releasing peptide receptor (GRPR) expression in prostate, breast, pancreatic and CNS tumours has driven the development of GRPR-targeted antagonists, including NeoBOMB1, RM2 and bombesin-derived analogues, because of the receptor’s mitogenic role in these malignancies.^{1 23} In comparative studies, ¹⁷⁷Lu-AMTG demonstrated modestly higher GRPR affinity, improved in vitro and in vivo stability and greater intratumoural retention at 24 hours than ¹⁷⁷Lu-RM2.²⁴ Emerging data suggest ¹⁶¹Tb-AMTG and its ¹⁷⁷Lu counterpart may provide superior therapeutic indices to RM2 constructs, with ¹⁶¹Tb-AMTG additionally benefiting from Auger electron emission.²⁵ NeoBOMB1 derivatives, including ⁶⁸Ga- and ¹⁷⁷Lu-labelled compounds, have also shown high GRPR-specific tumour uptake in PC-3 xenograft models.²⁶ Clinically, bombesin-derived ligands have shown promise in prostate cancer, where GRPR is frequently overexpressed.²⁷ Kurth et al reported the first dosimetric study of ¹⁷⁷Lu-RM2 in metastatic metastatic Castration Resistant Prostate Cancer (mCRPC) with low PSMA expression or exhausted standard options, demonstrating feasibility and tolerability.²⁸ The pancreas was identified as the dose-limiting organ, but rapid tracer clearance maintained absorbed doses within acceptable safety limits.²⁸

Human epidermal growth factor receptor 2

Human epidermal growth factor receptor 2 (HER2), a receptor tyrosine kinase within the EGFR family, drives oncogenic signalling and is most prominently overexpressed in breast carcinoma.²⁹ Positron emission tomography CT (PET/CT) evaluation of PEGylated liposomal ⁶⁴Cu-MM-302 demonstrated a 35-fold increase in tumour accumulation in HER2-overexpressing breast cancer, supporting its potential role in predicting therapeutic response.³⁰

The therapeutic evolution of HER2-targeted TRT is shifting from full-length monoclonal antibodies towards high-affinity, low-molecular-weight scaffolds such as nanobodies and peptides.¹ In a phase I study, ¹³¹I-GMIB-anti-HER2-VHH1 showed favourable biodistribution, manageable toxicity and significant lesion uptake.³¹ This radioiodinated single-domain antibody, developed for HER2-positive disease refractory to trastuzumab-based therapies, exerts its cytotoxicity primarily through β⁻ emission with a minor Auger electron contribution.³¹ Clinical dosimetry identified nephrotoxicity as the principal dose-limiting factor, observed only at higher administered activities.³¹

Integrins and immune-related targets (PD-L1)

Integrins αvβ3 and αvβ6 are key mediators of tumour proliferation and invasion and are overexpressed in a wide range of malignancies while remaining low or absent in most normal epithelial tissues, prompting early-phase clinical evaluation (NCT06375564).^{1 32}¹ Preclinical studies show that TRT can transiently upregulate programmed death-ligand 1 (PD-L1) expression within the tumour microenvironment and, when combined with anti–PD-L1 antibodies, significantly increase CD8⁺ T-cell infiltration, producing synergistic antitumour effects.^{33 34} Notably, concurrent administration appears superior to sequential scheduling in enhancing survival and achieving durable tumour control.³⁴

Clinically, radiolabelled arginine-glycine-aspartic acid analogues targeting αvβ3 provide a specific alternative to 2-deoxy-2-[¹⁸F]fluoro-D-glucose (¹⁸F-FDG)PET/CT in radioiodine-refractory thyroid cancer, particularly for cervical nodal and osteolytic metastases where FDG may be confounded by inflammation.³⁵

More broadly, integrating TRT with immune checkpoint inhibitors aims to harness radiation-induced immunogenic cell death and pro-inflammatory remodelling of the tumour microenvironment to elicit systemic antitumour immunity, including potential abscopal effects.³⁶ Historically, this concept was foreshadowed by the first Food and Drug Administration (FDA)-approved radioimmunotherapies, ⁹⁰Y-ibritumomab tiuxetan and ¹³¹I-tositumomab, for refractory follicular lymphoma, which demonstrated clinical efficacy but were ultimately limited by commercial and logistical factors, leading to withdrawal of ¹³¹I-tositumomab in 2014.³⁷ The expanding landscape of radiotheranostic targets and platforms is summarised in figure 1.

Figure 1. Radiotheranostics overview by molecular target, ligand class and representative radionuclides. The schematic links key oncologic targets, including FAP, CXCR4, GRPR, HER2, integrins and immune-related/CD20 targets, to corresponding ligand platforms (small molecules, peptides, antibody fragments and monoclonal antibodies) and representative diagnostic and therapeutic radionuclides. Colour coding distinguishes β-emitters, α-emitters, Auger/dual emitters and diagnostic tracers. CXCR4, chemokine receptor type 4; HER2, human epidermal growth factor receptor 2.

Novel radionuclides and emission paradigms

Key therapeutic radionuclides used in radiotheranostics differ substantially in emission characteristics, physical half-life and tissue penetration range, all of which influence therapeutic efficacy, toxicity profile, dosimetry and clinical applicability (table 1).

Table 1. Selected therapeutic radionuclides used in radiotheranostics and their key emission properties.

Radionuclide	Emission type	Physical half-life	Mean tissue range
90Y	β-emitter	64.1 hours (2.67 days)	~2.5–5 mm
177Lu	β-emitter with γ emission	6.65 days	~0.2–0.7 mm
161Tb	β-emitter with conversion/Auger electrons	6.9 days	~0.2–0.7 mm (β); nanometre scale for Auger electrons
225Ac	α-emitter	9.9 days	~50–100 µm
212Pb	In vivo α-generator/β-emitter	10.6 hours	~50–80 µm (via α-emitting daughters)
223Ra	α-emitter	11.4 days	~50–100 µm
211At	α-emitter	7.2 hours	~55–70 µm
227Th	α-emitter	18.7 days	~50–80 µm
213Bi	α-emitter	45.6 min	~50–80 µm

Next-generation β-emitters

Terbium-161 (¹⁶¹Tb) represents a next-generation β-emitter with physical properties similar to ¹⁷⁷Lu, but with the additional emission of low-energy conversion and Auger electrons that may enhance efficacy in small-volume or micrometastatic disease.³⁸ Early-phase clinical studies, including the VIOLET trial, are currently investigating ¹⁶¹Tb-labelled PSMA ligands in metastatic castration-resistant prostate cancer, based on preclinical evidence suggesting superior radiation delivery to micrometastatic disease compared with ¹⁷⁷Lu-labelled PSMA therapy, while maintaining a potentially acceptable safety profile (NCT05521412).³⁹ The SECuRE trial is also exploring ⁶⁷Cu-labelled agents as an alternative β-emitter (NCT04868604).

α-Particle therapy and high-LET radiotheranostics

Selection of α-emitters for TRT depends on half-life, decay properties, radiochemical stability and production feasibility, with key candidates including ²²⁵Ac, ²¹²Pb, ²²³Ra, ²¹¹At and ²²⁷Th.⁴⁰ These isotopes deliver high linear energy transfer (LET) over very short path lengths, enabling potent cytotoxicity while demanding high targeting precision.

²¹²Pb

The ^203/212Pb theranostic pair has emerged as a practical α-therapy platform, with first-in-human ²¹²Pb-DOTAMTATE in neuroendocrine tumours showing favourable pharmacokinetics, high tumour uptake, rapid renal clearance and measurable tumour reduction.^{41 42} These features support its further development as a next-generation somatostatin receptor (SSTR)-targeted radiotherapeutic.

²²³Ra

Radium-223 is the first α-emitting radiopharmaceutical to improve survival in bone-predominant mCRPC, delivering high-LET, short-range radiation that limits marrow toxicity and allows repeat dosing.⁴³ The ALpharadin in SYMptomatic Prostate CAncer (ALSYMPCA) trial established its clinical benefit, and subsequent studies, including RAdium LUtetium (RALU), have supported the safety of sequencing ²²³Ra with ¹⁷⁷Lu-PSMA therapy, while the ERA-223 trial highlighted the need for appropriate combination strategies with bone-protective agents.44,46

²¹¹At and ²²⁷Th

Astatine-211 and thorium-227 have shown early clinical and translational feasibility as antibody-based or small molecule-based α-emitters, expanding targeted α-therapy beyond bone and hematologic indications.4047,50

²²⁵Ac, ²¹³Bi and ²²⁷Th

In hematologic malignancies, ²¹³Bi- and ²²⁵Ac-lintuzumab target CD33, which is abundantly expressed on leukemic blasts but not stem cells, enabling selective cytotoxicity while sparing normal marrow.^{48 49} Extending this approach to lymphoid neoplasms, ²²⁷Th conjugated to anti-CD22 antibodies demonstrated first-in-human feasibility in lymphoma, underscoring the versatility of long-lived α-emitters in hematologic radiotheranostics.⁵⁰ Furthermore, in mCRPC utilising ²²⁵Ac alongside ¹⁷⁷Lu may bypass radio-resistance and optimise therapeutic efficacy; however, salivary gland toxicity, including xerostomia, remains a significant dose-limiting concern with alpha-emitter therapy.⁵¹

Radiobiological advantages and toxicity challenges

The paradigm of TRT provides a unique mechanism for overcoming tumour heterogeneity, a primary cause of failure in traditional oncology. A notable feature of beta-emitters, including the high-energy emitter ⁹⁰Y and lower energy emitter ¹⁷⁷Lu, is the ‘crossfire’ effect, whereby emitted radiation extends beyond the targeted cell to adjacent tissues. This may enhance treatment efficacy by irradiating neighbouring antigen-negative tumour cells and compensating for heterogeneous target expression; however, it may also contribute to unintended radiation exposure of surrounding normal tissues and organs. Conversely, while alpha-emitters and auger-emitting isotopes offer high LET, their clinical efficacy may be constrained by a lack of crossfire activity due to their nanometre-to-micrometre path lengths.⁴³

However, alpha emitters generate dense ionisation cascades culminating in DNA double-strand breaks, thereby conferring superior cytocidal efficacy against neoplastic cells relative to β-emitters.¹³ Tandem α-emitter and β-emitter therapy is a rational strategy that exploits their complementary radiobiologic properties to enhance tumour control while limiting organ toxicity.

Radiochemistry and ligand engineering innovations

Albumin binders and half-life extension strategies

The primary obstacle for FAPI radiolignad therapy (RLT) is rapid tumour washout, mismatching its physical and biological half-lives, leading to suboptimal radiation delivery and potential renal toxicity.^{52 53} To resolve this, albumin binders are incorporated to prolong circulation and increase cumulative tumour uptake.^{11 54 55} However, this can also increase off-target radiation to healthy tissues. Radionuclides with shorter half-lives, such as ⁶⁷Cu or ²¹¹At, better match biological residence time, improving therapeutic index, enhancing tumour dose, reducing normal tissue exposure and potentially shortening isolation requirements.⁵⁶

Improved chelators and linkers

The architectural configuration bridging a targeting ligand to a radioisotope determines metabolic stability, specificity and efficacy. Incorporating a linker between the peptide and chelator refines pharmacokinetics by improving hydrophilicity and biodistribution, while reducing steric hindrance to preserve receptor affinity. Amide linkers enhance stability and retention, whereas ester linkages allow enzyme-triggered release within the tumour microenvironment. Additionally, polyethylene glycol chains may be employed to increase aqueous solubility and circulatory half-life while attenuating immunogenicity.³⁶

Multivalent and bispecific radioligand

The pharmacological performance of FAPI ligands is largely determined by structural valency, which influences binding affinity and biological half-life. Early-generation monovalent ligands (eg, ¹⁷⁷Lu-FAPI-04 and -46) show rapid systemic clearance and favourable safety but are limited by short tumour residence times, often less than 2 hours, resulting in suboptimal tumour dose delivery. In contrast, bivalent constructs (eg, ¹⁷⁷Lu-DOTAGA) demonstrate ~eightfold higher affinity, tumour retention up to 168 hours and up to fourfold greater absorbed doses than monomers.¹²

Advanced designs, such as trivalent ligands (eg, ¹⁷⁷Lu-DOTA-FD2/3) and albumin-binding conjugates (eg, ¹⁷⁷Lu-EB-FAPI-B1 and FAPI-C16), prolong circulation and sustain high tumour uptake for 72–96 hours, thereby expanding the therapeutic window. However, extended systemic exposure increases the risk of hematologic toxicity, although slower clearance may reduce renal dose.¹²

Personalisation of radiotheranostics

From fixed-activity to dosimetry-guided therapy

Early clinical adoption of radiotheranostic therapies has largely relied on fixed administered activities, reflecting regulatory convenience and practical workflow considerations.⁵⁷ However, evidence from diverse clinical experiences consistently demonstrates substantial interpatient variability in radiopharmaceutical biodistribution, tumour uptake, clearance kinetics and organ-at-risk exposure.⁵⁸ In response to these observations, several regulatory bodies and professional societies, including the European Association of Nuclear Medicine (EANM), increasingly advocate and support the incorporation of dosimetry-guided radiopharmaceutical therapy into routine clinical practice.⁵⁹

Dosimetry-guided therapy individualises the administered activity according to patient-specific pharmacokinetics and predefined organ-dose and tumour-dose constraints, closely mirroring the principles of treatment planning in external beam radiotherapy.⁶⁰

A cornerstone of dosimetry-guided therapy is the reduction of dose-limiting toxicity to healthy organs. EANM emphasises the use of patient-specific absorbed-dose estimates to maintain organ-at-risk exposure within accepted safety thresholds, particularly for the kidneys and bone marrow, rather than relying on fixed administered activities.² This principle is supported by the VISION dosimetry substudy, which demonstrated that cumulative renal doses were generally maintained below ~23 Gy and marrow doses below ~2 Gy across six cycles of ¹⁷⁷Lu-PSMA therapy, with higher rates of clinically relevant toxicity observed in patients approaching or exceeding these absorbed-dose ranges.⁶¹ Collectively, this individualised approach improves organ-at-risk protection while preserving therapeutic efficacy.

In addition to guiding adherence to safety constraints, dosimetry plays a critical role in predicting and monitoring treatment response by quantifying the absorbed dose delivered to tumour lesions. Multiple studies have shown that responders to ¹⁷⁷Lu-PSMA therapy receive substantially higher absorbed doses in lymph node and bone metastases than non-responders, with nearly twofold higher doses per administered activity in lymph node lesions (3.73 vs 1.86 Gy/GBq) and bone lesions (3.47 vs 1.48 Gy/GBq).⁶² Prospective investigations further demonstrate that median whole-body tumour absorbed doses of ~11.55 Gy correlate with biochemical response, with PSA declines ≥50% observed at higher tumour doses (14.1 Gy vs 9.6 Gy), while meaningful responses are uncommon below a ~10 Gy threshold.⁶³ Serial quantitative single-photon emission computed tomography/computed tomography (SPECT/CT) analyses reinforce these findings, linking high cumulative tumour absorbed doses to marked tumour volume reduction and sustained PSA decline, with projected doses exceeding 40 Gy across multiple cycles in selected responders.⁶⁴

Despite these promising results, there is no randomised evidence that dosimetry-guided personalisation improves overall survival over fixed-activity ¹⁷⁷Lu-PSMA therapy; current data mainly support gains in safety and response prediction rather than survival. The ongoing PROstate-specific Membrane Antigen DosImetry-Guided EndoradiotherapY (PRODIGY-1) Phase I/II trial (NCT05896371) is prospectively evaluating a kidney-dose-guided strategy with individualised administered activities, including dose escalation to define renal safety limits, extended treatment cycles and dosimetry-guided salvage therapy, while assessing toxicity, response, progression-free survival and overall survival. Complementary early-phase clinical studies, systematic reviews and emerging reduced-burden and automated dosimetry workflows further support the feasibility and scalability of personalised dosing approaches in clinical practice.^{65 66}

Lesion-based and voxel-level dosimetry

Lesion-based dosimetry enables correlation of absorbed dose with lesion-specific response, revealing that individual metastases within the same patient may receive markedly different doses and exhibit discordant responses, as demonstrated in ¹⁷⁷Lu-PSMA and ¹⁷⁷Lu-DOTATATE therapies.⁶⁷ In contrast, voxel-level dosimetry estimates absorbed dose at the level of individual imaging voxels (≈2–5 mm), generating three-dimensional dose maps that capture spatial heterogeneity in radiation delivery. Using quantitative PET/SPECT with Monte Carlo or kernel methods, voxel-level dosimetry more accurately models radiation transport than mean dose estimates; integrated with SPECT/CT, it enables personalised therapy by capturing heterogeneous activity distribution, with recent computational advances improving clinical feasibility.⁶⁸

Recent improvements in dosimetry protocols

Recent advances in ¹⁷⁷Lu-PSMA dosimetry have focused on simplifying imaging protocols to reduce patient burden and resource demands while maintaining acceptable accuracy. Single time-point dosimetry at 48–72 hours can estimate absorbed doses to normal organs within ~±20% of multitime-point (MTP) reference methods, allowing faster assessment with fewer scans and reduced inconvenience for patients, although tumour dose estimates remain less reliable due to heterogeneous uptake and clearance.⁶⁹ Dual time-point protocols offer improved lesion dosimetry, particularly when a late acquisition (eg, 72–168 hours) is included; appropriate late-time point pairing can reduce lesion dose uncertainty to ~13% and closely approximate full MTP accuracy, enabling more precise lesion-level treatment planning while minimising additional imaging sessions.⁷⁰ Beyond post-therapy imaging, the ‘first-strike’ dosimetry concept integrates pretherapy PSMA PET with rapid calculation tools to estimate tumour and organ doses before treatment, enabling upfront activity personalisation.⁷¹ Early clinical data show a tumour dose–response relationship, with higher absorbed doses linked to greater PSA declines, supporting simplified biology-driven dosimetry for personalisation lesion. However, dosimetry remains limited by heterogeneous uptake, variable clearance and partial-volume effects inadequately captured by simplified imaging.

Artificial intelligence and digital theranostics

AI-assisted segmentation and quantification

Artificial intelligence has fundamentally reshaped image-based tumour segmentation in both PET/CT and SPECT/CT, enabling automated, reproducible and high-throughput quantification of tumour burden.⁷² Deep-learning-based algorithms, particularly convolutional neural networks, can automatically delineate tumour volumes across whole-body datasets, facilitating objective calculation of volumetric biomarkers. These parameters provide a more comprehensive representation of total disease burden than conventional SUV-based metrics, particularly in patients with multifocal or disseminated disease.⁷³ By replacing manual or semiautomated contouring, AI can enhance time reduction and limit operator dependency.⁷⁴

Within the theranostic paradigm, automated segmentation plays a pivotal role in longitudinal treatment assessment and dosimetry-guided therapy. AI-based models, including U-Net-derived architectures, have demonstrated high spatial concordance with expert-defined gross tumour volumes on PSMA and somatostatin receptor-targeted imaging, enabling consistent lesion tracking across treatment cycles of ¹⁷⁷Lu-PSMA and ¹⁷⁷Lu-DOTATATE therapies.⁷⁵ This consistency is crucial for accurately assessing therapy-induced changes in tumour burden, estimating absorbed dose and classifying response. Moreover, by standardising tumour delineation across time points, AI facilitates reliable comparison of baseline and posttherapy imaging, supporting early response evaluation and personalised treatment adaptation. Collectively, these capabilities position AI-driven segmentation as a foundational component of digital theranostics, bridging molecular imaging, quantitative biomarkers and individualised radionuclide therapy planning.⁷⁵

AI in dosimetry and outcome prediction

Beyond absorbed dose estimation, AI-driven radiomics and dosiomics are increasingly explored for advanced response and dose–effect modelling.⁷⁶ Radiomic features derived from baseline PET/CT, including texture, heterogeneity and intensity metrics, have been associated with treatment response and survival and, when combined with dosimetric data, can reflect relationships between image phenotype and absorbed dose.⁷⁷ When integrated with voxel-level dose distributions (dosiomics), these features characterise spatial dose heterogeneity within tumours and organs, capturing information not represented by mean dose alone.⁷⁸

Across peptide receptor radionuclide therapy (PRRT) and PSMA radioligand therapy, multimodal models consistently outperform single-modality imaging. In neuroendocrine tumours, machine learning approaches using CT or SPECT/SPECT-CT radiomics achieve response-prediction accuracies of approximately 70%–89%, generally exceeding SPECT alone.⁷⁹ Multimodal fusion of ⁶⁸Ga-DOTATATE PET/CT with clinical and laboratory variables further improves progression-free survival risk stratification, although most evidence remains retrospective and requires external validation.⁸⁰ A similar pattern is observed in prostate cancer, where models integrating PSMA PET with additional tracers and clinical or biological markers better predict response to ¹⁷⁷Lu-PSMA, underscoring that outcome modelling in radiotheranostics is inherently multidomain rather than imaging-only.⁸¹

Dosimetry-enabled and AI-enabled workflows show feasibility and links to safety and response, but their true impact requires prospective validation and practical integration into routine care. Transitioning from fixed-activity dosing to quantitative personalisation—paralleling the evolution of external beam radiotherapy—could, if widely adopted, transform radioligand therapy into an adaptive, data-driven treatment model. Workforce limitations also extend to radiochemists, who are essential for radiopharmaceutical manufacturing and quality assurance and are increasingly in short supply globally.

Combination strategies: radiotheranostics plus immunotherapy

Radiotheranostics and immunotherapy are complementary pillars of precision oncology, and growing evidence suggests that their integration can yield synergistic effects.⁸² TRT delivers ionising radiation to tumour cells and the tumour microenvironment, producing not only direct cytotoxicity but also immunomodulatory effects that may enhance responsiveness to immune-based treatments.⁸³ Ionising radiation is increasingly recognised as an inducer of immunogenic cell death, promoting the release of damage-associated molecular patterns such as calreticulin, high-mobility group box 1 (HMGB1), and adenosine triphosphate (ATP) which facilitate dendritic cell activation and antigen presentation.^{83 84} In addition, the low-dose, protracted radiation typical of β-emitting radionuclides can modulate the tumour microenvironment by increasing MHC-I expression, enhancing antigen presentation and shifting cytokine profiles towards antitumour immunity.⁸⁵

Unlike external beam radiotherapy, radiotheranostics provide systemic, molecularly targeted irradiation and can potentially prime multiple metastatic sites simultaneously.⁸² This has fuelled interest in combining TRT with immune checkpoint inhibitors. Preclinical data indicate that TRT can reduce immunosuppressive components of the tumour microenvironment, including regulatory T cells and myeloid-derived suppressor cells, while promoting effector T-cell infiltration, supporting a role for radiotheranostics as immune sensitisers rather than purely cytotoxic agents.^{82 83}

Emerging clinical evidence supports the feasibility of these combinations, particularly in prostate cancer and neuroendocrine tumours. In the PRINCE trial, ¹⁷⁷Lu-PSMA-617 plus pembrolizumab achieved high PSA50 response rates with an acceptable safety profile in men with mCRPC, prompting further studies exploring sequencing and priming strategies, often incorporating PSMA PET for lesion-level assessment.⁸⁶ Similarly, ¹⁷⁷Lu-DOTATATE combined with nivolumab has shown tolerability and preliminary activity in pulmonary neuroendocrine tumours and small cell lung cancer.⁸⁷ These studies signal a shift from purely cytotoxic objectives toward exploiting radiopharmaceuticals as systemic immune modulators, although randomised trials are still needed to define survival benefit, optimal sequencing and patient selection.

The current wave of development reflects a maturation of radiotheranostics from a niche therapy into a platform embedded within multimodal oncology care, where the key challenge is no longer efficacy in selected indications but rational integration, personalisation and sequencing across disease stages and biological subtypes, supported by trials that incorporate imaging, dosimetry and molecular endpoints alongside conventional outcomes.

Clinical trials, regulatory landscape and implementation

Ongoing and landmark clinical trials

Radiotheranostics is rapidly transitioning from early-phase development to practice-changing trials, particularly in prostate cancer and neuroendocrine tumours.⁵¹ In prostate cancer, the phase III VISION, phase II TheraP, and phase III PSMAfore trials demonstrated improved outcomes with ¹⁷⁷Lu-PSMA-617 compared with standard-of-care options, supporting regulatory approval and progressive expansion of indications, figure 2.688,90 In neuroendocrine tumours, the NETTER-1 and NETTER-P trials established the role of ¹⁷⁷Lu-DOTATATE, leading to FDA and European Medicines Agency (EMA) approval for gastroenteropancreatic NETs in adults and adolescents.^{7 91 92} Earlier milestones, including the ALSYMPCA trial of radium-223 in metastatic prostate cancer and studies of ¹³¹I-MIBG in pheochromocytoma/paraganglioma, remain foundational to current practice.⁴⁴

Figure 2. Evolution of ¹⁷⁷Lu-PSMA radioligand therapy clinical trials (2021–2025). ARPI, Androgen receptor pathway inhibitor; FDA, Food and Drug Administration; mCRPC, metastatic Castration Resistant Prostate Cancer; mHSPC, metastatic hormone-sensitive prostate cancer; PCa, prostate cancer; PSMA, prostate-specific membrane antigen; RCT, randomised controlled trial.

More recent studies have focused on refining indications and moving ¹⁷⁷Lu-PSMA earlier in the disease course, figure 3. The phase II LUNAR trial showed improved progression-free survival with neoadjuvant ¹⁷⁷Lu-PSMA-I&T plus stereotactic body radiotherapy compared with radiotherapy alone in oligorecurrent hormone-sensitive prostate cancer, without added toxicity.⁹³ The phase II ENZA-p trial demonstrated improved PSA progression-free survival with ¹⁷⁷Lu-PSMA-617 plus enzalutamide in ADT-progressed disease, and the UpFrontPSMA trial reported higher PSA response rates when ¹⁷⁷Lu-PSMA-617 was combined with docetaxel in de novo high-volume metastatic hormone-sensitive prostate cancer.^{94 95} The ongoing phase III PSMAddition trial is evaluating even earlier use of ¹⁷⁷Lu-PSMA-617 in combination with hormonal therapy (NCT04720157). In parallel, the phase III NETTER-2 trial demonstrated improved progression-free survival when ¹⁷⁷Lu-DOTATATE was added to long-acting octreotide in patients with grade 2–3 advanced GEP-NETs, extending the NETTER-1 paradigm.⁹⁶

Figure 3. Combination strategies with ¹⁷⁷Lu-PSMA radioligand therapy across therapeutic domains. Schematic overview illustrating clinical trial–based combinations of ¹⁷⁷Lu-PSMA with hormone therapy, chemotherapy, immunotherapy, PARP (poly(ADP-ribose) polymerase) inhibition, other radioligands, surgery and external radiation. PSMA, prostate-specific membrane antigen; SABR, stereotactic ablative body radiotherapy.

Although ¹⁷⁷Lu-based theranostics currently dominate clinical practice, multiple alternative radionuclide–ligand pairings are under active investigation. These include ¹⁷⁷Lu-edotreotide (¹⁷⁷Lu-ITM-11) for GEP-NETs in the phase III COMPOSE trial (NCT04919226), ¹⁶¹Tb-PSMA-I&T in the phase I/II VIOLET trial for mCRPC (NCT05521412), and several early- to late-phase studies evaluating α-emitters such as ²²⁵Ac and ²¹²Pb.⁵¹ In this context, ²²⁵Ac-PSMA-617 is being explored as a post–¹⁷⁷Lu-PSMA option in mCRPC in the phase II/III PSMAcTion trial (NCT06780670), while α-emitter strategies are also extending to neuroendocrine tumours, including ²¹²Pb-VMT-α-NET in SSTR2-positive disease (NCT05636618).

Beyond prostate cancer and NETs, a broad early-phase pipeline targets additional malignancies. Examples include ¹⁷⁷Lu-PentixaTher for CXCR4-positive relapsed/refractory acute leukaemia (PENTILULA; NCT06356922), ¹⁷⁷Lu-DOTATATE in recurrent meningioma (MOMENTUM-1; NCT06955169), ²²⁵Ac-DOTATATE in SSTR-positive breast cancer (TRACY-1; NCT06590857), and ²²⁵Ac-SSO110 for small cell lung cancer and Merkel cell carcinoma (SANTANA-225; NCT06939036). FAP-directed approaches are also advancing, including ¹⁷⁷Lu-FAP-2286 (LuMIERE; NCT04939610), ²¹²Pb-PSV359 (NCT06710756), and ²²⁵Ac-RTX-2358 (ATLAS; NCT07156565). Early studies are further exploring radioimmunotherapy constructs targeting PD-L1 and HER2 (¹⁷⁷Lu-RAD204, NCT06305962; ¹⁷⁷Lu-RAD202, NCT06824155) and nanoparticle-based delivery systems such as ¹⁷⁷Lu-SN201 (NCT06184035). Most of these agents are being developed with matched diagnostic counterparts, and ongoing and planned phase II–III trials are summarised in table 2.

Table 2. Summary of radiotheranostic pairs under evaluation in current or planned phase II or III clinical trials.

Biological target (studied malignancy)	Radiotheranostic pair (trial phase)	Trial identifier(s)
PSMA (prostate cancer)	177Lu- & 68Ga/18F-labelled agents (Phase III)	NCT06520345, NCT06496581, NCT06320067, NCT05939414, NCT05204927, NCT04720157, NCT04689828, NCT04647526, NCT04876651, NCT00859781
	225Ac & 68Ga/18F-labelled agents (Phase III)	NCT06855277, NCT06780670
	161Tb- & 68Ga/18F-labelled agents (Phase I/II)	NCT05521412, NCT07208240
	212Pb-ADVC001 & 68Ga/18F-labelled agents (phase I/II)	NCT05720130
SSTR (neuroendocrine tumours, meningioma, breast cancer, Merkle cell carcinoma)	177Lu- & 68Ga-labelled SSTR agonists (Phase III)	NCT05918302, NCT05387603, NCT04919226, NCT03049189, NCT04261855, NCT06955169
	225Ac-DOTATATE (RYZ101) & SSTR PET agent (Phase I/II & III)	NCT05477576, NCT07150806, NCT06590857
	212Pb- & 202Pb-VMT-alpha-NET (Phase I/II)	NCT06427798, NCT05636618
	177Lu-DOTA-JR11 & SSTR PET agent (Phase I/II)	NCT04997317
	225Ac-SSO110 & SSTR PET agent (Phase I/II)	NCT06939036
FAP (solid tumours)	177Lu- & 68Ga-FAPI-2286 (Phase I/II)	NCT04939610
	212Pb- & 203Pb-PSV359 (Phase I/II)	NCT06710756
CXCR4 (acute leukaemia, CNS lymphoma)	177Lu-/90Y-PentixaTher & 68Ga-PentixaFor (Phase I/II)	NCT06356922, NCT06132737
GRPR (breast cancer)	177Lu- & 68Ga-NeoB (phase I/II)	NCT06247995
CA-IX (solid tumours including renal carcinoma)	177Lu- & 68Ga-DPI-4452 (Phase I/II)	NCT05706129
	177Lu- & 89Zr-girentuximab (Phase I/II)	NCT05663710, NCT05239533
CD276 or B7-H3 (solid tumours)	177Lu-BetaBart (Phase I/II)	NCT07189871
	131I-omburtamab (phase II)	NCT04022213, NCT04743661
MC1R (melanoma)	212Pb-VMT01 & 203Pb/68Ga-VMT02 (Phase I/II)	NCT05655312
CD25 (Hodgkin lymphoma)	90Y-labelled Anti-CD25 monoclonal antibody (Phase II)	NCT04871607
CD45 (diffuse large B-cell lymphoma)	131I-apamistamab (Phase I/II)	NCT06768905
LAT1 (glioblastoma)	131I-iodofalan & 18F-FET (Phase III)	NCT07100730
Norepinephrine transporter (neuroblastoma, pheochromocytoma and paraganglioma)	131I- & 123I-MIBG (Phase II)	NCT00107289

225Ac, Actinium-225; ARPI, Androgen receptor pathway inhibitor; CA-IX, carbonic anhydrase IX; CD, cluster of differentiation; CXCR4, C-X-C motif chemokine receptor type 4; 18F, Fluorine-18; FAP, fibroblast activation protein; FDA, Food and Drug Administration; 68Ga, Gallium-68; GRPR, gastrin-releasing peptide receptor; 131I, Iodine-131; LAT1, L-type amino acid transporter 1; 177Lu, Lutetium-177; MC1R, melanocortin 1 receptor; mCRPC, metastatic castration-resistant prostate cancer; mHSPC, metastatic hormone-sensitive prostate cancer; 212/203Pb, Lead-212/203; PCa, prostate cancer; PSMA, prostate-specific membrane antigen; RCT, randomised controlled trial; SSTR, somatostatin receptor; 161Tb, Terbium-161; 90Y, Yittrium-90; 89Zr, Zirconium-89.

Regulatory challenges

The rapid expansion of radiotheranostics poses distinct challenges for regulators, who must balance patient safety with timely clinical translation. Because radiotheranostics lie at the interface of pharmaceutical and radiation regulation, authorisation is often split across multiple agencies, creating additional procedural barriers and delaying patient access, particularly at the local level. A more centralised or coordinated regulatory framework, or early parallel mapping of drug and radiation requirements by developers, could streamline translation from preclinical to clinical stages.

There is also increasing emphasis on patient-specific dosimetry to optimise safety and efficacy. Both the FDA and EMA require detailed dosimetric data for marketing authorisation, which can complicate and prolong early-phase trials and increase development costs.⁹⁷ Guidance from professional societies such as the EANM aims to standardise and simplify dosimetry workflows in radiotheranostic trials, and further integration of AI may reduce the time and resources needed for reliable dose estimation.^{75 98}

Finally, regulatory harmonisation remains fragmented across jurisdictions, with substantial variation in requirements and clinical practice. This heterogeneity impedes both research and access. International and regional initiatives, including those led by the International Atomic Energy Agency (IAEA) and the Australasian Radiopharmaceutical Trials Network (ARTnet), seek to promote standardisation and mitigate the negative impact of regulatory fragmentation on clinical development.⁹⁹

Global access and health-system readiness

Beyond regulatory hurdles, health-system readiness represents a major determinant of equitable access to radiotheranostics. Increasing technical complexity and cost create substantial barriers to implementation, particularly in resource-limited settings. Addressing these constraints is essential if radiotheranostics is to evolve from a highly specialised service into a broadly accessible component of cancer care.

Infrastructure and training

Centres without prior radiotheranostic experience must invest in advanced radiochemistry facilities, PET/SPECT imaging and radiation-safe therapy suites, alongside specialised radiation protection measures such as automated dispensing, shielding, contamination monitoring and tailored waste-management protocols.^{100 101} These requirements impose significant infrastructural demands and operational responsibilities. In parallel, there is a growing shortage of trained radiopharmacists, medical physicists and nuclear medicine physicians with radiotheranostic expertise. Without a skilled workforce and robust quality systems, even approved therapies cannot be delivered safely or at scale.¹⁰² International bodies, including the IAEA, emphasise strengthening education and integrating radiotheranostics into formal training curricula to build a sustainable workforce.^{100 103}

Cost, affordability and supply chains

Radiotheranostics remain resource-intensive to develop and deliver, and the combined costs of radiopharmaceuticals and specialised infrastructure are major barriers, especially in low and middle-income countries.¹⁰⁴ Supply chain limitations further exacerbate inequities. Many key radionuclides are produced in ageing reactors or specialised cyclotrons, and only a minority of global facilities have sufficient production capacity to meet growing demand.^{101 104} This concentration of supply increases vulnerability to shortages, as illustrated historically by disruptions in Mo-99/⁹⁹ᵐTc, and more recently by constraints affecting α-emitter availability.105,107 In addition, the relatively short physical half-lives of several clinically relevant radioisotopes impose major logistical and distribution challenges, limiting transportation range, scheduling flexibility and timely delivery to treatment centres.¹⁰¹ Efforts to streamline logistics, decentralise production where feasible and strengthen international coordination are therefore critical to building resilient supply chains.^{108 109}

Radiotheranostics’ future impact will depend as much on health-system readiness as scientific innovation, as gaps in regulation, workforce, manufacturing and reimbursement may restrict access to specialised centres. Ensuring equitable and sustainable access should, therefore, be viewed as a core component of the radiotheranostic agenda rather than a secondary implementation challenge.

Conclusion

Radiotheranostics is rapidly evolving through advances in target biology, radionuclide diversity, molecular engineering and quantitative, digitally guided planning. The coming decade will hinge not on a single breakthrough, but on integrating four pillars: biologically informed targeting, emission-matched and chemically optimised design, dosimetry-guided and AI-guided personalisation and scalable health-system infrastructure.

Clinically, the shift must move from empiric algorithms to data-driven strategies defining patient selection, agent choice, dosing and sequencing. System-level progress in regulation, manufacturing, training and reimbursement will determine whether radiotheranostics becomes widely accessible or remains confined to specialised centres.

Coordinated scientific and infrastructural advances could transform radiotheranostics into a scalable precision oncology platform integrating molecular imaging, TRT and digital decision support for personalised care.

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study.