Combining Diagnostic Imaging and Pathology for Improving Diagnosis and Prognosis of Cancer

Orazio Schillaci, Manuel Scimeca, Nicola Toschi, Rita Bonfiglio, Nicoletta Urbano, Elena BonannoContrast 2019 · DOI

TL;DR

This review paper advocates for a synergistic approach between diagnostic imaging (like PET-CT and MRI) and digital pathology to improve cancer diagnosis and prognosis. By integrating molecular insights from imaging with detailed cellular and genetic information from pathology, clinicians can achieve a more personalized and effective treatment strategy for cancer patients.

What Problem Does This Paper Address?

The management of cancer patients is increasingly moving towards personalized medicine, which demands a deep, multi-faceted understanding of each individual's disease. Current diagnostic methods, while advanced, often operate in silos. Diagnostic imaging techniques, such as PET-CT and MRI, excel at detecting cancer at a molecular level, identifying metabolic changes and tumor kinetics early on, even before macroscopic tumors form. However, they provide a more macroscopic view of the disease. On the other hand, histopathology, the examination of tissue samples, offers crucial, detailed information about the cellular and molecular characteristics of the tumor, including genetic alterations and protein expression. The challenge lies in effectively integrating these two powerful but distinct diagnostic modalities to provide a comprehensive picture that can guide prognosis and therapy. This paper addresses the need for a translational and multidisciplinary approach that bridges the gap between these fields.

Methods

This paper is a review article, so it doesn't present novel experimental methods. Instead, it synthesizes existing research and discusses the integration of established and emerging techniques. The core "methods" discussed involve:

  • Molecular Imaging: Techniques like Positron Emission Tomography (PET) and Computed Tomography (CT) (PET-CT), Magnetic Resonance Imaging (MRI), and Single-Photon Emission Computed Tomography (SPECT) are highlighted for their ability to detect molecular kinetics, angiogenesis, cell metabolism, and receptor expression in vivo. The use of specific radiotracers (e.g., 18F-FDG, 18F-choline, anti-HER2 antibodies) is central to this approach.
  • Anatomic Pathology: This includes traditional histopathology (H&E staining) and advanced ancillary techniques like immunohistochemistry (IHC) for identifying biomarkers (e.g., ER, PR, Ki67, HER2, PSA, PSMA).
  • Digital Pathology: The paper emphasizes the transformative impact of Whole Slide Imaging (WSI) and digital slide scanners. This technology allows for the digitization of entire pathology slides, enabling remote access, advanced image analysis, and integration with other data sources.
  • Computer-Aided Diagnosis (CAD) and Artificial Intelligence (AI): The authors discuss the potential of AI, particularly Deep Neural Networks (DNNs), to analyze complex multimodal data from both imaging and digital pathology. DNNs can identify subtle patterns, predict molecular alterations, and potentially aid in diagnosis, prognosis, and treatment response prediction.

The paper advocates for the combination of these methods, suggesting that molecular imaging can guide where to biopsy, and pathology can validate and further characterize findings from imaging.

Datasets

Not discussed in the paper. The paper is a review and does not present original experimental data or introduce new datasets.

Key Insights and Ideas

  • The "One-Stop Shop" Concept: Combined imaging techniques like PET-CT offer the potential for a single examination to provide comprehensive diagnostic information, reducing patient burden and time to diagnosis.
  • Bridging the Spatial and Temporal Gap: Biopsies provide a snapshot of a tumor at a specific time and location, but tumors are heterogeneous and evolve. Molecular imaging can offer a more dynamic, whole-body view, complementing the detailed cellular information from pathology.
  • Digital Pathology as an Enabler: WSI is revolutionizing pathology by making slides accessible for advanced computational analysis and facilitating telepathology and collaborative diagnostics.
  • AI as a Unifier: DNNs have the potential to integrate vast amounts of data from imaging, digital pathology, and omics technologies, leading to more accurate and personalized diagnoses.
  • Therapeutic Guidance: The integration of imaging and pathology can directly inform treatment decisions, such as selecting patients for targeted therapies (e.g., anti-HER2 antibodies) or predicting response to immunotherapies.
  • Microcalcifications as a Biomarker: The paper highlights research showing that microcalcifications in breast cancer, when analyzed both radiologically and histologically, can provide prognostic information and are linked to specific cellular processes.

Key Results

As a review paper, it doesn't present quantitative results from its own experiments. Instead, it summarizes findings from cited studies:

  • Improved Staging: PET-CT has been shown to be superior to PET or CT alone for TNM staging of tumors.
  • Predictive MRI Features: Studies have identified MRI features that can predict molecular patterns in breast cancer (e.g., ER, PR, HER2 status).
  • Correlation of Imaging and Pathology: Several studies demonstrate correlations between molecular imaging uptake (e.g., 18F-FDG, 99mTc sestamibi) and histopathological markers (e.g., Ki67, ER, PR, HER2) in various cancers.
  • Potential for Early Detection: Techniques like 99mTc sestamibi SPECT show promise for early detection of breast cancer lesions with high metastatic potential.
  • AI's Diagnostic Power: DNNs have demonstrated high success rates in oncology for tasks like classifying breast lesions, predicting survival, and identifying metastases from medical images.

Limitations and Future Work

  • Standardization of Digital Pathology: The operational phases of digital pathology are not yet fully standardized, which is a prerequisite for widespread clinical adoption.
  • Data Requirements for AI: DNNs require large, manually curated datasets for effective training, necessitating efforts in multicentric data collection and standardization.
  • Correlation Challenges: While correlations exist, exact associations between certain imaging parameters (like FDG uptake) and specific pathological markers (like Ki67) are still being refined.
  • Future Directions: The paper strongly advocates for continued collaboration between diagnostic imaging and anatomic pathology departments. Future work should focus on developing AI algorithms that can integrate multimodal data for improved diagnosis, prognosis, and personalized treatment strategies. The development of new molecular imaging tracers and advanced digital pathology analysis tools will also be crucial. The ultimate goal is to create a synergic framework that truly serves personalized medicine.

Summary updated

Abstract

In the era of personalized medicine, the management of oncological patients requires a translational and multidisciplinary approach. During early phases of cancer development, biochemical alterations of cell metabolism occur much before the formation of detectable tumour masses. Current molecular imaging techniques, targeted to the study of molecular kinetics, employ molecular tracers capable of detecting cancer lesions with both high sensitivity and specificity while also providing essential information for both prognosis and therapy. On the contrary, complementary and crucial information is provided by histopathological examination and ancillary techniques such as immunohistochemistry. Thus, the successful collaboration between diagnostic imaging and anatomic pathology can represent a fundamental step in the "tortuous" but decisive path towards personalized medicine.
BibTeX
@inproceedings{schillaci2019combining,
  author = {Orazio Schillaci and Manuel Scimeca and Nicola Toschi and Rita Bonfiglio and Nicoletta Urbano and Elena Bonanno},
  title = {{Combining Diagnostic Imaging and Pathology for Improving Diagnosis and Prognosis of Cancer}},
  booktitle = {Contrast},
  year = {2019},
  doi = {10.1155/2019/9429761},
}
Figure