Nomograms for predicting long-term survival in patients with non-metastatic nasopharyngeal carcinoma in an endemic area.

PURPOSE

Nomogram for predicting more than a 5-year survival for non-metastatic nasopharyngeal carcinoma (NPC) was lacking. This study aimed to develop the new nomograms to predict long-term survival in these patients.

RESULTS

The median follow-up time for training set and test set was 95.2 months and 133.3 months, respectively. The significant predictors for death were age, gender, body mass index (BMI), T stage, N stage, lactate dehydrogenase (LDH), and radiotherapy techniques. For predicting recurrence, age, gender, T stage, LDH, and radiotherapy techniques were significant predictors, whereas age, gender, BMI, T stage, N stage and LDH were significant predictors for distant metastasis. The calibration curves showed the good agreements between nomogram-predicted and actual survival. The c-indices for predicting death, recurrence, and distant metastases between nomograms and the TNM staging system were 0.767 VS.0.686 (P<0.001), 0.655 VS.0.585 (P<0.001), and 0.881 VS.0.754 (P<0.001), respectively. These results were further confirmed in the test set.

METHODS

On the basis of a retrospective study of 1593 patients (training set) who received radiotherapy alone or concurrent chemoradiotherapy from 2000 to 2004, significant predictors were identified and incorporated to build the nomograms. The calibration curves of nomogram-predicted survival versus the actual survival were plotted and reviewed. Bootstrap validation was performed to calculate the concordance index (c-index). These models were further validated in an independent prospective trial (test set, n=400).

CONCLUSION

The established nomograms suggest more-accurate long-term prediction for patients with non-metastatic NPC.