Related Experiment Video
Updated: May 11, 2026
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
Prognostic nomogram for diffuse large B-cell lymphoma incorporating the International Prognostic Index with
1Haematology Department, Sir Charles Gairdner Hospital, Perth, Western Australia, Australia.
Background/Aims:
Results from interim-positron emission tomography (PET) studies in diffuse large B-cell lymphoma (DLBCL) patients are varied. We evaluated the prognostic value of interim-PET in our centre. To improve concordance, interim-PET was combined with the International Prognostic Index (IPI).
Methods:
We retrospectively reviewed 100 new consecutive DLBCL patients treated with immunochemotherapy from 2005 to 2010. Twenty-four patients did not receive interim-PET and were excluded. Interim-PET images were re-examined using a qualitative assessment technique. Progression-free survival (PFS) and overall survival (OS) were analysed by the Cox proportional hazards model and prognostic accuracy was assessed using Harrell's C statistics (C).
Results:
Eleven patients were positive, and 65 were negative at interim-PET. The 2-year OS and PFS were 70.8% and 60.0%, respectively, in the PET-negative group, 36.4% and 36.4% for the PET-positive group (log-rank P-value 0.0008 for PFS, 0.0001 for OS). The IPI and interim-PET were minimally correlated. On Cox regression analysis, both were significant indicators of PFS (P < 0.001 and P = 0.002 respectively). The prognostic accuracy for PFS of a negative PET result was limited (C = 0.63), as it was for IPI (C = 0.75), but with the two indicators combined, the predictive accuracy was improved (C = 0.81). A nomogram, predictive for relapse-free survival at 2 years, was constructed.
Conclusion:
In DLBCL patients treated with immunochemotherapy, the IPI and interim-PET provide independent prognostic information. In combination, a more powerful predictive model may be created as a nomogram. This can be refined in prospective trials and may help clinical decision making.