The 2024 European Society for Medical Oncology (ESMO) Annual Meeting was successfully held in Barcelona, Spain. At this conference, a study led by Dr. Guodong He from Zhongshan Hospital, utilizing the PDNet model to predict the efficacy of chemoradiotherapy combined with immunotherapy in MSS locally advanced rectal cancer patients, was selected for poster presentation (Abstract No. 551P). During the event, Oncology Frontier had the opportunity to speak with Dr. He, who provided an in-depth interpretation of the study’s findings.________________________________________
Oncology Frontier: Immunotherapy has advanced rapidly in recent years, making the selection of patients who will benefit from treatment a key clinical challenge. Your ESMO-selected study explored the use of the PDNet model to predict the efficacy of chemoradiotherapy combined with immunotherapy in MSS locally advanced rectal cancer. Could you introduce the background of this research?
Dr. Guodong He: Our study primarily focuses on mid-to-low rectal cancer, which has long posed a challenge in surgical and comprehensive treatment. The complexity of treating this type of cancer goes beyond oncological outcomes—it also involves considerations of organ function and organ preservation. The ultimate goal is to ensure optimal treatment efficacy while maintaining quality of life for patients.
Existing data suggest that immunotherapy combined with radiotherapy can significantly improve clinical outcomes in MSS rectal cancer patients. There have been numerous studies in China exploring this combination, and several reports have already been published. However, many unanswered questions remain, particularly in identifying which patients will benefit the most from immunotherapy. Key considerations include the presence of predictive biomarkers, changes in the immune microenvironment, and the long-term impact of early tumor regression following radiotherapy and immunotherapy.
With these gaps in knowledge, our research aimed to leverage artificial intelligence (AI) to analyze MRI images and predict which patients are most likely to benefit from chemoradiotherapy combined with immunotherapy.
________________________________________
Oncology Frontier: What were the key findings of your study?
Dr. Guodong He: Our study enrolled patients with locally advanced mid-to-low rectal cancer, all of whom underwent long-course radiotherapy, followed by 2–3 cycles of anti-PD-1/PD-L1 therapy, and radical resection. MRI images were collected before radiotherapy, and for each patient, three images with the clearest tumor visualization were selected and annotated.
We developed the PDNet model to extract features from these MRI images and predict whether patients would achieve pathological complete response (pCR). A total of 61 patients were included, all of whom achieved 100% R0 resection. Among them, 24 patients (39.3%) achieved pCR.
The study used a dataset of 183 rectal MRI images, with 147 images used for training and 36 images for testing. The PDNet model demonstrated high accuracy in internal validation, with a classification accuracy of 86.11%, a positive predictive value of 93.33%, and a negative predictive value of 80.95%. The model’s sensitivity was 77.78%, while specificity reached 94.44%. Statistical analysis showed a p-value of 0.0000498 and an area under the curve (AUC) of 0.8580 (95% CI: 0.6982–0.9805).
In summary, the study demonstrated a strong correlation between MRI-extracted features and the likelihood of achieving pCR. Using the PDNet model, we can predict a patient’s probability of achieving pCR before undergoing chemoradiotherapy combined with PD-1/PD-L1 inhibitors. We are currently conducting external validation of the model.
________________________________________
Oncology Frontier: What are the implications of this study for clinical decision-making in MSS locally advanced rectal cancer?
Dr. Guodong He: At this year’s ESMO conference, both domestic and international studies have increasingly focused on immunotherapy. These studies are not only concerned with whether patients achieve pCR, but also explore molecular biomarkers such as circulating tumor DNA (ctDNA) to assess treatment responses and underlying mechanisms.
Our research follows a similar approach. By using the PDNet model to identify patients most likely to benefit from immunotherapy, we can implement more personalized treatment strategies. Patients identified as responders may have the option to avoid surgery, preserve organ function, and ultimately improve their quality of life.
Conversely, if the model predicts that a patient is unlikely to achieve pCR, a watch-and-wait strategy may not be suitable. In such cases, clinicians may need to consider surgical intervention or close monitoring. I believe this model provides valuable guidance for oncologists in designing individualized treatment plans. With the accumulation of more research data, this model has the potential to further refine patient selection for immunotherapy, ultimately benefiting more patients in the future.
________________________________________
Oncology Frontier: What were some of the most impressive research advancements at this year’s ESMO conference?
Dr. Guodong He: One of the most notable developments was the innovative treatment strategies for advanced MSS colorectal cancer. In the era of precision medicine, treatment for advanced colorectal cancer is no longer a one-size-fits-all approach where chemotherapy is the default first-line therapy. Instead, patient selection based on gene expression profiles is becoming increasingly important, guiding the use of immunotherapy and targeted therapy. Some studies are now exploring the integration of targeted therapy, immunotherapy, and chemotherapy as first-line treatment for advanced colorectal cancer.
Another key advancement was in dMMR/MSI-H colorectal cancer. Over the past few years, immune checkpoint inhibitors (ICIs) have demonstrated promising efficacy in dMMR/MSI-H colorectal cancer, showing significant potential. The NICHE-2 study revealed that neoadjuvant nivolumab plus ipilimumab led to substantial improvements in major pathological response (MPR) and pCR rates.
At this year’s ESMO conference, new findings from the NICHE-3 study were presented, investigating nivolumab plus relatlimab in locally advanced, resectable dMMR colorectal cancer. The results were equally encouraging. Both NICHE-2 and NICHE-3 have now been accepted by Nature Medicine, highlighting their clinical significance. These findings are expected to influence future treatment guidelines and bring meaningful changes to clinical practice.
