Innovation thrives on necessity.
Nuclear physics, biotechnology and clinical practice have come a long way, and in the era of AI, these advances are leading to innovative solutions that are truly impacting lives. In the last two decades we have seen tremendous advancement in therapies for cancer in particular. Still, a remarkable proportion of patients with solid or rare tumors continue to have limited treatment options, particularly in the advanced stages of disease. Apart from surgery, chemotherapy and radiation therapy remain the mainstays of cancer therapy, however their non-specificity to tumor cells, and associated systemic toxicity poses a need for more targeted and better tolerated approaches.
Radiopharmaceutical Therapy (RPT), also referred to as targeted radionuclide therapy or radioligand therapy (RLT), is a fast advancing and highly promising innovation in the oncology treatment landscape. In RPT, the radioactive isotope is linked to a special molecule, called a ligand, that is designed to seek out and attach to specific markers (a target protein) found on the surface of cancer cells. The isotope emits localised high energy radiation ( alpha, beta particles) that destroys the cancer cell. Both the specificity and abundance of targets on the cancer cells determine the success of RPT. For example, PSMA (Prostate-Specific Membrane Antigen) is a common target for prostate cancer and its metastasis, while SSTR (Somatostatin Receptor) is a target for the neuroendocrine tumors commonly found in the stomach, intestine, and pancreas. This precise and targeted approach in RPT spares the adjacent healthy tissue causing less side effects and better quality of life for the patient.
The Mechanics of a More Connected Healthcare Ecosystem [Video]
Arbiter’s Anjali Jameson on hospital and payer alignment.
With regulatory approvals accelerating, new radiotracers and radiopharmaceutical therapies are being used in earlier stages of disease in the clinical practice. While the field is moving from research and innovation to application in mainstream oncology, it still falls short of its potential. For example, a significant share of prostate cancer patients on PSMA-targeted RPT such as Pluvicto – so called “non-responders” – gain minimal or no measurable benefit, yet many continue therapy for months because early non-response is difficult to detect with current response evaluation framework, creating a “non-responder gap.”
This “non-responder gap” has both clinical and economic implications. Patients who meet today’s selection or eligibility criteria for RPT show a heterogeneous treatment response, meaning results can vary widely. This makes it difficult to predict which patients will actually respond to RPT. In prostate cancer RPT, published trials and real-world series show wide heterogeneity in response: PSA50 (≥50% decline in Prostate Specific Antigen level) responder rates cluster around ~30–50% and many patients have minimal or no biochemical/radiographic response. In a recently published study, over 80% of mCRPC-advanced prostate cancer patients discontinued RPT-¹⁷⁷Lu-PSMA-617, driven by progression and toxicity. Discontinuation independently worsens survival, emphasizing the need for careful biomarker-guided patient selection for RPT to improve treatment adherence and outcomes. Similarly, In Neuroendocrine tumor RPT, published trials showed a modest objective response rate of 18%, in a subset of NET patients. Typically, RPTs are administered using “one-size-fits-all” dosing schedules over multiple cycles, often without a standardized framework for early response assessment. As a result, patients who are primary or early non-responders will receive additional treatment cycles before changes in therapy are considered. This demonstrates that in real world clinical workflows, clinicians face significant uncertainty in selecting the right therapy at the right time for the right patient.
Non-responder gap: Why it matters
The non-responder gap matters for two reasons: First, patients who are in an advanced stage of cancer lose valuable time in ineffective therapy. Second, it increases the cost of therapy and resources for both patients and the broader healthcare ecosystem.
The Hidden Administrative Tasks Draining Small Practices
Small practices play a critical role in healthcare delivery, but they cannot continue to absorb ever-increasing administrative demands without consequences.
For patients, continuing ineffective RPT can delay the transition to alternative options such as chemotherapy, other targeted agents, clinical trials, or combination approaches — interventions that may offer earlier disease control. Additional RPT cycles also unnecessarily increase cumulative radiation exposure and potential toxicities without clinical benefit.
Given the high per-patient cost of RPT and the limited high resource infrastructure required to deliver them, unnecessary treatment cycles represent a significant source of waste. The cost of futile therapies is borne not only by hospitals and payers, but also by the patients themselves. Even when institutions are able to cover the majority of costs for RPT, patients are often responsible for substantial out-of-pocket costs. In many cases, patients will even travel to different states to obtain treatment. Clearly, early identification of non-responders is key, and it can be achieved through the use of tools and methods which enhance risk stratification and optimize treatment decisions by untapping the potential of quantitative imaging.
Yet, limited resources, clinician burnout, variability between readers, scanners and lack of standardized protocols all make it difficult to gain reliable insight and inflate the nonresponder gap across the patient population.
Bridging this gap requires not just better drugs and improved access; it requires optimized use of information for better decisions. That’s where AI can be used innovatively and impactfully.
AI as a clinical companion
Closing the non-responder gap will require tools that can learn, scale, and convert imaging data into actionable insight irrespective of individual experience. AI can be most impactful if utilized to reduce the non-responder gap by two interventions: Firstly, AI can untap the potential of imaging information to identify potential non-responders before even starting therapy. Secondly, it can identify non-responders effectively as early as the completion of the first treatment cycle, saving valuable time for patients and enhancing clinician confidence through data-driven decision-making.
RPT can be used when a tumor has expression of target proteins which is detected by radioligand tracer (RLT) imaging. In many clinical workflows, imaging is still only used qualitatively and patient selection for RPT is based on manual visual assessment for the presence or absence of radioligand positive lesions. A clear illustration of this is RLT imaging like PSMA PET/CT, SSTR PET/CT, which are used to detect expression of target proteins and have become central to patient selection for RPT treatment planning in prostate cancer and neuroendocrine tumors, respectively.
Baseline RLT imaging contains valuable information about how a patient’s disease is likely to respond to RPT. Tumor burden, lesion heterogeneity, tracer uptake patterns, and spatial distribution all carry predictive value. However, extracting this information consistently and at scale is difficult in routine clinical practice.
AI addresses this challenge by automating quantitative image analysis, which includes segmentation, measurement, and characterization of disease burden. In PSMA radioligand imaging, this means generating objective measurements such as total tumor volume, target expression intensity, disease heterogeneity, and absorbed dose across organs and lesions.
Instead of relying on limited manual image analysis, AI is able to detect subtle imaging changes and heterogeneous expressions that might otherwise be missed, and combine them with clinical data to predict or flag early signs of resistance or non-response. These explainable, objective, and reproducible biomarkers can reduce reliance on subjective interpretation and minimize inter-reader variability. In doing so, they create a more reliable path toward personalised cancer treatment.
The role of AI in this setting is not to replace the physician’s judgement, but to strengthen it. When these insights become an integral part of clinical workflow, imaging becomes a dynamic decision tool, reducing uncertainty for clinicians.
As new radiotracers and radiopharmaceutical therapies like alpha particle therapies and novel target RPTs (fibroblast activated protein or FAP) are introduced, clinicians will continue to face evolving interpretive guidelines. Different tracers demonstrate distinct biodistribution and clinicians will need the ability to accurately interpret complex images, differentiate physiological and abnormal uptake, and avoid false positives. This creates a steep learning curve to gain specific knowledge about how these compounds behave in the body. AI systems trained across diverse datasets can help normalize this complexity and reduce knowledge gap errors. These AI platforms can provide more consistent-tracer agnostic- assessments, reducing discrepancies and leading to more standardized and reliable reporting for patient selection.
Scaling radiopharmaceutical therapy responsibly and effectively will depend not only on new drugs but also on smarter interpretation of the data already available. Using AI to close the non-responder gap is how precision oncology becomes both more effective and sustainable.
Photo: Love Employee, Getty Images
Jayashri Pawar, MD serves as Medical Director at Nucs AI, where she leads the company’s vision and AI strategy to advance precision imaging in prostate cancer and oncology care. She brings deep clinical insight to the responsible development and deployment of AI technologies, ensuring accuracy, quality, and meaningful impact on patient outcomes. With more than 15 years of experience in diagnostic medical imaging, Dr. Pawar has broad expertise across molecular imaging, CT, MRI, ultrasound, and emergency radiology, paired with hands-on leadership in building and scaling AI/ML and generative AI solutions in healthcare.
Previously, she served as Medical Director of Clinical AI at Covera Health, helping shape enterprise AI strategy and supporting national deployment of computer vision, NLP, and GenAI models for radiology quality and population health. She completed a Clinical Innovation Fellowship at Massachusetts General Hospital and Brigham and Women’s Hospital’s Center for Clinical Data Science, advancing AI solutions toward FDA clearance and clinical adoption.
Dr. Pawar is also an active member of the Point-of-Care Ultrasound (POCUS) community and is committed to bridging clinical practice, innovation, and AI leadership to help define the future of imaging-driven care.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.