Enhancing Outcome Prediction in Cervical Cancer: A Multivariable Approach with Hypoxia-related Parameters

Article Outline

• Purpose/Objective(s)
• Materials/Methods
• Results
• Conclusions
• Copyright

Back to Article Outline

Purpose/Objective(s) 

Accurate prediction of tumor response and therapy outcome can play a critical role in cancer care. Hemoglobin (Hgb) levels, tumor volume and perfusion status obtained during early therapy were reported to independently influence therapy outcome in advanced cervical cancer. However, the overall predictive accuracy remains to be improved. This study was to determine if a comprehensive approach by augmenting these 3 factors can further improve predictive power for ultimate treatment outcome in cervical cancer.

Back to Article Outline

Materials/Methods 

MRI imaging based tumor volume, perfusion, and hemoglobin levels were analyzed in 88 cervical cancer patients (stage IB2-IVA), who were part of a serial MR imaging protocol. Patients were treated with standard RT/chemotherapy. MRI imaging obtained in early-RT (20-24 Gy/2-2.5 weeks) included 3D tumor volume measurement on T₂-weighted imaging, and tumor perfusion analysis with dynamic-contrast-enhanced (DCE) MRI. Tumor perfusion was quantified by the previously established DCE parameter lowest 10th percentile of signal intensity (SI10), representing poorly-perfused pixel populations within the tumor. Individual and combined parameters were correlated with local (pelvic) tumor control and disease-free survival (follow-up: 0.1-9.0 years, mean 4.6 years).

Back to Article Outline

Results 

All 3 parameters in early RT significantly and independently correlated with therapy outcome. The sensitivity, specificity and accuracy in predicting local control was 80%, 73%, and 74% for Hgb alone, 80%, 67%, and 69% for volume alone, 80%, 53%, and 58% for perfusion alone, and improved to 80%, 86%, and 85% for the combination of all 3 parameters. Specifically, the combination of low-Hgb (<12.2 g/dl), low-perfusion (SI10 <2.1), and large-volume (≥40 cm³) resulted in a 5-year local control rate of only 41%, compared to 95% for high Hgb, high perfusion, and smaller volume (p < 0.001). Disease-free survival was 36% and 74%, respectively (p < 0.001). The combined parameters also improved the sensitivity, specificity and accuracy in predicting disease-free survival from 50%, 70%, and 64% for Hgb alone, 67%, 72%, and 70% for volume alone, and 73%, 44%, and 55% for perfusion alone to 48%, 86%, and 74% for the combination of all 3 parameters.

Back to Article Outline

Conclusions 

A multimodality approach for outcome prediction by laboratory and imaging parameters can further improve accuracy over the use of individual parameters alone. These results suggest that noninvasive measurement of parameters related to absolute oxygen content per unit tumor tissue, as reflected by the efficacy in oxygen delivery (perfusion status and Hgb level), and normalized by the amount of tumor tissue (volume), further improves the predictive power for ultimate treatment outcome.