Implementation of Precision Oncology into the Diagnostic and Therapeutic Armamentarium: Actionable Takeaways from the 2023 Precision Oncology Summit

Molecular Imaging in Precision Medicine

Molecular imaging is a rapidly advancing field that holds great promise in cancer research, diagnosis, and treatment. It integrates advanced imaging technologies with the principles of cellular and molecular biology, allowing researchers and clinicians to visualize and understand the molecular processes associated with cancer.¹

A. Nuclear Medicine Modalities Used in Oncology

Nuclear medicine encompasses a variety of imaging modalities that utilize radioactive materials administered in trace amounts (radiotracers) to visualize and assess the function of organs and tissues.² The primary nuclear medicine modalities include²:

• Single-Photon Emission Computed Tomography (SPECT): SPECT is a nuclear imaging technique that uses gamma-ray-emitting radiotracers to create three-dimensional images of the distribution of the radiotracer within the body.

• Positron Emission Tomography (PET): PET is an imaging technique that involves the use of positron-emitting radiotracers. When a positron collides with an electron, it produces two gamma-ray photons, which are detected by a PET scanner. PET provides information about metabolic and biochemical processes. Common radiotracers include Fluorine-18 (F-18) FDG for glucose metabolism and Gallium-68 (68Ga) PSMA for prostate cancer imaging.

• SPECT/Computed Tomography (SPECT/CT): SPECT/CT combines the functional information from SPECT with the anatomical detail obtained from CT imaging. This fusion of functional and structural data enhances the accuracy of localization and improves the interpretation of nuclear medicine studies.

• PET/CT: PET/CT integrates the metabolic information from PET with the anatomical detail provided by CT. This combination allows for precise localization of functional abnormalities, aiding in the diagnosis and staging of various diseases, including cancer.

• PET/Magnetic Resonance Imaging (PET/MRI): PET/MRI is a hybrid imaging modality that combines the metabolic information of PET with the detailed soft tissue contrast of MRI.

• Theragnostics: Theragnostics involves the use of radiotracers for both diagnostic imaging and targeted radionuclide therapy. For example, the FDA-approved Lutetium-177 (Lu-177) PSMA617 is the targeted radionuclide arm of the PSMA-based theragnostics used in prostate cancer. The diagnostic arm includes the FDA-approved ⁶⁸Ga-PSMA11 and ¹⁸F-DCFPyL.

B. Prostate-Specific Membrane Antigen (PSMA)-PET Scan

The approval and implementation of PSMA-PET scan has highlighted the advancement of molecular imaging for patients with prostate cancer. PSMA is a transmembrane protein overexpressed in prostate cancer with known enzymatic activities that acts as glutamate-preferring carboxypeptidase.¹ The expression of PSMA is significantly increased in prostate cancer cells. This overexpression makes PSMA an attractive target for diagnostic imaging and therapy in prostate cancer. Imaging modalities such as PET or SPECT use radiotracers that specifically bind to PSMA, allowing for the visualization of prostate cancer lesions in the body. Various radiotracers have been developed for PSMA-targeted imaging, including those labeled with 68Ga or F-18 for PET imaging. PSMA-PET has shown high sensitivity in detecting prostate cancer lesions, including primary tumors and metastases.³ In addition to diagnostic imaging, PSMA is a target for radionuclide therapy in prostate cancer. Radiolabeled compounds targeting PSMA, such Lu-177 PSMA617, are used for targeted radionuclide therapy. This approach, known as theragnostics, allows for both diagnosis (imaging) and therapy using the same target.⁴

C. Clinical applications of PSMA-PET in prostate cancer

PSMA imaging and therapy have become important tools in the management of prostate cancer. PSMA-PET is used for staging of high-risk prostate cancer and evaluation at biochemical recurrence. PSMA-targeted radionuclide therapy is employed in selected cases, especially in patients with advanced disease. However, while PSMA-targeted imaging has shown excellent sensitivity, it is not entirely specific to prostate cancer, and PSMA expression can also be found in some non-cancerous tissues. Interpretation of imaging results often requires correlation with clinical context and other diagnostic information. Ongoing research aims to improve the performance of PSMA-targeted imaging and therapy. This includes the development of new radiotracers, optimization of imaging protocols, and investigations into the use of PSMA in different stages of prostate cancer. Other emerging applications of PSMA-targeted imaging and therapy include the identification of oligometastatic disease, guidance of focal therapy, and the assessment of treatment response.^5,6

⁶⁸Ga-PSMA11 was FDA approved for PET imaging of PSMA-positive lesions in patients with prostate cancer in 2020.⁷ Later in 2021, the CONDOR and OSPREY trials led to the U.S. Food and Drug Administration (FDA) approval of ¹⁸F-DCFPyL (Pylarify^®, Lantheus Holdings, Inc., USA) as the first PSMA-PET imaging agent available nation-wide for prostate cancer.⁸

The CONDOR study (NCT03739684)⁹ was designed to evaluate the performance of 18F-DCFPyL-PET/CT in patients who had biochemical recurrence and uninformative standard imaging results. It enrolled men with rising prostate specific antigen (PSA) ≥ 0.2 ng/mL after prostatectomy or ≥ 2 ng/mL above nadir after radiotherapy. The primary outcome was correct localization rate (CLR), defined as positive predictive value with an additional requirement of anatomic lesion colocalization between ¹⁸F-DCFPyL-PET/CT and a composite standard of truth (SOT). The SOT consisted of, in descending priority (i) histopathology, (ii) subsequent correlative imaging findings, or (iii) post-radiation PSA response. A total of 208 men underwent ¹⁸F-DCFPyL-PET/CT. The CLR was 84.8%-87.0% (lower bound of 95% confidence interval [CI]: 77.8-80.4). The disease detection rate was 59% to 66% (at least one lesion detected per patient by ¹⁸F-DCFPyL-PET/CT by central readers). As the study’s primary endpoint, the ¹⁸F-DCFPyL-PET/CT provided clinically meaningful and actionable information about disease localization despite negative standard imaging.⁹ The phase 2/3 OSPREY trial (NCT02981368) divided patients into 2 cohorts, with cohort A including patients with high-risk, locally advanced prostate cancer, and cohort B including patients with metastatic or recurrent disease. This study showed improvements in the specificity (96%-99%) and positive predictive value (78%-91%) of ¹⁸F-DCFPyL when compared with conventional imaging for metastatic prostate cancer.¹⁰ Both trials led to an approval of PSMA-PET imaging agent that is a game-change for men facing prostate cancer.

In summary, the theragnostic approach offers the potential for personalized and targeted treatment strategies, with prostate cancer being one of the pioneers in this area. Furthermore, the combination of molecular imaging with omics technologies and analyses of circulating tumor cells seems necessary to gain a better understanding of tumor biology and develop novel therapeutic strategies.¹¹

Machine-Learning in Precision Oncology

Machine-learning approaches have indeed emerged as powerful tools in cancer research and treatment. These approaches leverage computational methods to analyze large datasets, identify patterns, and extract valuable insights from complex biological and clinical information. In the context of cancer, machine learning holds great potential for identifying treatment responses, understanding pathways, and improving outcomes, particularly in cases with poor prognosis.¹²

Machine learning models can analyze clinical and molecular data to predict how an individual patient may respond to a specific treatment. This personalized approach helps in tailoring treatment plans based on the unique characteristics of each patient’s cancer. Machine learning algorithms can also identify novel biomarkers associated with poor prognosis or treatment resistance. By analyzing diverse data types, including genomics, proteomics, and imaging data, these algorithms contribute to the discovery of new targets for intervention. In addition, machine learning techniques enable the identification of distinct subtypes within a particular cancer type. This molecular subtyping allows for more precise patient stratification, helping to identify those at higher risk or those who may benefit from specific therapies. It can uncover complex non-linear relationships within biological pathways and networks. This knowledge is valuable for understanding the underlying mechanisms of cancer development, progression, and resistance to treatment.^12–15

Moreover, machine learning can be applied to radiomics, which involves extracting quantitative features from medical images. This approach aids in characterizing tumors, predicting treatment response, and assessing prognosis based on imaging data. Integrating diverse omics data, such as genomics, transcriptomics, proteomics, and metabolomics, is a complex task. Therefore, the use of machine learning helps integrate and analyze these datasets comprehensively, revealing intricate relationships and providing a holistic view of cancer biology.^14,15 This is also used in drug discovery by predicting potential therapeutic targets and identifying existing drugs that may be repurposed for cancer treatment, including for metastatic prostate cancer. This accelerates the drug development process and increases the chances of identifying effective therapies.^13,15 For example in one study of metastatic prostate cancer patients followed for nine years who were profiled for hundreds of circulating plasma-based lipid species and circulating tumor DNA (ctDNA) somatic alterations, the use of a ML approach incorporating multi-omic features was found to significantly improve the prediction of metastatic prostate cancer outcomes, thereby providing valuable support for informed decision-making and efficient modeling for future patient.¹⁶

The integration of machine learning algorithms into the clinical practice can assist clinicians in making real-time decisions by analyzing patient data during treatment. This dynamic approach allows for adaptive treatment strategies based on evolving patient characteristics. This also contributes to the development of robust prognostic and diagnostic models. These models use clinical and molecular data to predict patient outcomes, recurrence risk, and aid in early cancer detection.¹² Furthermore, using a machine learning approach could assists in identifying patient subgroups that are more likely to respond to specific treatments. This can improve the success rates of clinical trials by enrolling patients who are more likely to benefit.¹⁷

While machine learning holds great promise, its successful application requires rigorous validation, integration with traditional research methods, and careful consideration of ethical and interpretability aspects. As technology and methods continue to evolve, machine learning is likely to play an increasingly integral role in advancing our understanding of cancer and improving patient outcomes.

Conflict of Interest

Dr. Andrei Iagaru: Alpha9Tx - Scientific Advisory Board; Clarity Pharmaceuticals - Scientific Advisory Board; GE Healthcare - research grant; Novartis Pharmaceuticals - research grant, consulting, study steering committee; Progenics Pharmaceuticals – consulting; Radionetics Oncology - Scientific Advisory Board; RayzeBio – Consulting; Telix – Consulting.

The other authors do not have a conflict of interest.

Funding Information

N/A

Ethical Statements

N/A

Acknowledgement

The authors thank the Binaytara Foundation for the opportunity to highlight this important topic.

Author Contributions

• All authors: conception and design.

• All authors: data collection and assembly.

• IR and IA: data analysis and manuscript writing.

All authors have approved this manuscript.