Development and Validation of Nomograms Predictive of Overall and Progression-Free Survival in Patients With Oropharyngeal Cancer

Abstract

Purpose Treatment of oropharyngeal squamous cell carcinoma (OPSCC) is evolving toward risk-based modification of therapeutic intensity, which requires patient-specific estimates of overall survival (OS) and progression-free survival (PFS). Methods To develop and validate nomograms for OS and PFS, we used a derivation cohort of 493 patients with OPSCC with known p16 tumor status (surrogate of human papillomavirus) and cigarette smoking history (pack-years) randomly assigned to clinical trials using platinum-based chemoradiotherapy (NRG Oncology Radiation Therapy Oncology Group [RTOG] 0129 and 0522). Nomograms were created from Cox models and internally validated by use of bootstrap and cross-validation. Model discrimination was measured by calibration plots and the concordance index. Nomograms were externally validated in a cohort of 153 patients with OPSCC randomly assigned to a third trial, NRG Oncology RTOG 9003. Results Both models included age, Zubrod performance status, pack-years, education, p16 status, and T and N stage; the OS model also included anemia and age × pack-years interaction; and the PFS model also included marital status, weight loss, and p16 × Zubrod interaction. Predictions correlated well with observed 2-year and 5-year outcomes. The uncorrected concordance index was 0.76 (95% CI, 0.72 to 0.80) for OS and 0.70 (95% CI, 0.66 to 0.74) for PFS, and bias-corrected indices were similar. In the validation set, OS and PFS models were well calibrated, and OS and PFS were significantly different across tertiles of nomogram scores (log-rank P = .003;< .001). Conclusion The validated nomograms provided useful prediction of OS and PFS for patients with OPSCC treated with primary radiation-based therapy.

Fig 1.

Overall survival (OS). (A) Calibration plot of OS at 5 years. Nomogram-predicted OS is plotted on the x-axis, with observed OS on the y-axis. Dashed lines along the 45-degree line through the origin point represent the perfect calibration models in which the predicted probabilities are identical to the actual probabilities. (B) Nomogram for predicting probability of OS at 2 and 5 years. The presence or absence of each clinical characteristic indicates a certain number of points. Number of points for each clinical characteristic is on the top row. For each characteristic, the absence is assigned zero points. The presence of characteristics is associated with number of points. The points for each characteristic are summed together to generate a total-points score. The total points correspond to respective 2-year and 5-year OS probabilities. (C) Predicted 2-year (thick dark line with short dashed 95% CIs) and 5-year (thin dark line with long dashed 95% CIs) OS probability on the basis of nomogram total points.

Fig 2.

Progression-free survival (PFS). (A) Calibration plot of PFS at 5 years. Nomogram-predicted PFS is plotted on the x-axis, with observed PFS on the y-axis. Dashed lines along the 45-degree line through the origin point represent the perfect calibration models in which the predicted probabilities are identical to the actual probabilities. (B) Nomogram for predicting probability of PFS at 2 and 5 years. The presence or absence of each clinical characteristic indicates a certain number of points. Number of points for each clinical characteristic is on the top row. For each characteristic, the absence is assigned zero points. The presence of characteristics is associated with number of points. The points for each characteristic are summed together to generate a total-points score. The total points correspond to respective 2-year and 5-year PFS probabilities. (C) Predicted 2-year (thick dark line with short dashed 95% CIs) and 5-year (thin dark line with long dashed 95% CIs) PFS probability on the basis of nomogram total points. PS, performance status.

Fig 3.

Survival curves for validation cohort. (A) Overall survival (OS) curves for the validation cohort for the three groups defined by tertiles of the total points from the derivation cohort (P values: first tertile versus second tertile, .05; second tertile versus third tertile, .04). (B) Progression-free survival curves for the validation cohort for the three groups defined by tertiles of the total points from the derivation cohort (P values: first tertile versus second tertile, < .001; second tertile versus third tertile, .3).

Fig A1.

CONSORT diagram for Radiation Therapy Oncology Group (RTOG) 0129 and RTOG 0522.

Fig A2.

CONSORT diagram for Radiation Therapy Oncology Group (RTOG) 9003.