TL;DR
This pilot study explores the use of confocal laser endomicroscopy (CLE) during surgery for oropharyngeal squamous cell carcinoma (OPSCC) to assess tumor margins in real-time. The findings suggest CLE can be integrated into the operating room, providing crucial microscopic views that could lead to more precise tumor removal and potentially less radical surgery.
What Problem Does This Paper Address?
Surgical removal of OPSCC aims to achieve "safe margins" – ensuring all cancerous tissue is excised while preserving as much healthy tissue as possible. Traditionally, this relies on macroscopic assessment and later histopathological analysis of excised tissue. However, determining these margins accurately during surgery can be challenging, potentially leading to either incomplete tumor removal (requiring further surgery or adjuvant treatment) or overly aggressive resection, impacting patient quality of life. This paper investigates whether CLE, a real-time microscopic imaging technique, can provide immediate feedback to surgeons about the presence of cancer at the surgical margin, thereby improving the precision of tumor resection.
Methods
The researchers conducted a prospective pilot study involving five patients undergoing OPSCC resection. During surgery, they used a probe-based CLE system, which employs a laser to generate high-resolution, in-vivo microscopic images of tissue. After injecting fluorescein dye (which highlights intercellular spaces and microvasculature), the CLE probe was used to image both the tumor and the surrounding healthy-appearing mucosa. These CLE images were then compared to the gold standard of hematoxylin and eosin (H&E) stained tissue samples. A key aspect of the study was a blinded assessment of 169 representative CLE images by four medical professionals (three head and neck surgeons and one pathologist) who were tasked with classifying them as either "healthy" or "carcinoma."
Datasets
The study did not utilize pre-existing datasets for training or evaluation. Instead, it generated its own dataset of 12,809 CLE frames from five patients, from which 169 representative images were selected for analysis. These images were then classified by blinded examiners against the histopathological findings of the corresponding tissue samples.
Key Insights and Ideas
- Real-time microscopic "biopsy": CLE offers a way to obtain "optical biopsies" in real-time, allowing surgeons to visualize cellular structures and tissue architecture directly during the operation, without waiting for traditional pathology reports.
- Distinct visual signatures: Healthy oropharyngeal tissue exhibits a regular, organized cellular structure with clear cell borders and normal vascular patterns. In contrast, OPSCC shows a disorganized cellular arrangement, irregular cell shapes, and abnormal microvasculature, which are visible with CLE.
- Potential for less radical surgery: By providing immediate feedback on margin status, CLE could help surgeons make more informed decisions about the extent of resection, potentially avoiding unnecessary removal of healthy tissue and preserving function.
- Integration into workflow: The study demonstrated that CLE could be easily integrated into the intraoperative setting, adding only a small amount of time to the overall procedure.
Key Results
The blinded assessment of CLE images by the medical professionals yielded promising results. The study reported an overall accuracy of 85.8% for classifying tissue as cancerous or healthy. Specific metrics included:
- Sensitivity: 89.9% (correctly identifying cancerous tissue)
- Specificity: 78.6% (correctly identifying healthy tissue)
- Positive Predictive Value (PPV): 87.9%
- Negative Predictive Value (NPV): 81.9%
The inter-rater reliability, measured by Fleiss' kappa, was 0.60, indicating a "moderate" to "substantial" level of agreement among the examiners.
Limitations and Future Work
The authors acknowledge several limitations:
- Pilot study size: The study involved only five patients, which is a small sample size.
- Inter-rater variability: While moderate agreement was achieved, there was still some inconsistency among the examiners, particularly between surgeons with different levels of experience with CLE.
- Distinguishing CIS from invasive carcinoma: The maximum penetration depth of the CLE probe (approximately 60 µm) makes it difficult to differentiate between carcinoma in situ and invasive carcinoma, as stromal invasion cannot be assessed.
- Cost and learning curve: The high cost of CLE equipment and the learning curve associated with image interpretation are significant barriers to widespread clinical adoption.
Future work should focus on larger, multi-center studies to validate these findings. Developing automated classification algorithms for CLE images could also help overcome inter-rater variability and reduce the reliance on expert interpretation. Further research is needed to establish the clinical utility and cost-effectiveness of CLE in OPSCC surgery.
Summary updated