Clinical and biological features predictive of survival after relapse of neuroblastoma: A study from the International Neuroblastoma (NB) Risk Group (INRG) Database.

Abstract

9518 Background: In NB, most patients (pts) who relapse eventually die. Prognostic factors are used to stratify treatment at diagnosis, but typically not at the time of relapse. Our goals were to determine a) which factors were predictive of time to death post-relapse; b) if time from diagnosis until relapse/progression has a predictive role. Methods: Retrospective analysis included INRG pts with first event of relapse, progressive disease, or secondary malignancy (excluding pts whose first event was death). Time from diagnosis until event (“time-to-first-event”) was calculated and analyzed as <1 year (yr) vs ≥1 yr. 5-yr estimates of overall survival (OS ± standard error), time from first event until death or last contact, are presented (lifetable methods). Time-to-first-event was tested in a multivariable Cox model (adjusting for nonproportional hazards) with clinical and biologic factors; hazard ratios (HR) for increased risk of death post-relapse were calculated. Results: From 8,800 INRG pts, 2,266 experienced a non-death first event. Median time to relapse was 13.2 months (mo) (range: 1 day to 11.4 yrs). The 5-yr OS after first event was 20%±1%. Time-to-first-event (HR=1.8), age >18 mo (HR=2.3), INSS stage 4 (HR=3.4), MYCN amplified (HR=2.8), diploidy (HR=1.6), high MKI (HR=2.0), undifferentiated grade (HR=1.6), and 1p aberration (HR=1.7) were significantly predictive of death after relapse (p<0.0001), but not 11q aberration. Compared to pts whose first event occurred <6 mo from diagnosis, pts who relapsed 6-<18 mo from diagnosis had increased risk of death, while relapses ≥18 mo from diagnosis had decreased risk of death. Shorter time- to-first-event was not independently predictive of death after adjustment for undifferentiated grade, high MKI, MYCN amplification, or diploidy. Conclusions: Time to first relapse is a significant predictor of time to death after relapse; however, the risk of death is higher for pts who relapse within 6-<18 mo, but lower for pts who relapse ≥18 mo from diagnosis. Stratification of relapsed NB pts according to the timing of first relapse, age, stage, MYCN, and MKI, and diploidy is important in retrieval study designs. No significant financial relationships to disclose.