Abstract 1110
Background
Serum AFP levels are used as a diagnostic and prognostic marker for pts with HCC. Assessment of the clinical relevance of changes in AFP over time outside of clinical trials is lacking. This study was designed to explore the relationship between changes in AFP levels and overall survival (OS) in a real-world setting.
Methods
This retrospective analysis used the IMS Oncology EMR database (US patients, 12/01/2007-12/31/2014). Eligible pts were diagnosed with HCC, 18+ years old, had at ≥ 1 AFP test recorded 60 days prior to 180 days after diagnosis, and received anticancer therapy ≤180 days after diagnosis. Survival analyses were by Kaplan-Meier method. The gamma-frailty model was used to correlate AFP change utilizing previously-reported definitions of AFP change (above/below 400 ng/mL, ≥20 ng/mL, ≥7 ng/mL/month, ±20%, ±50%) and OS.
Results
A total of 907 pts met eligibility criteria (77.3% male, median 65 years of age). Of 697 pts with AFP prior to start of first-line therapy, the 453 (65%) with a baseline AFP <400 ng/mL had an OS of 4.2 months and the 244 (35%) with ≥400 ng/mL an OS of 2.9 months. An increase in AFP was associated with a decrease in OS in the 278 patients with baseline and first-line therapy (1L) AFP (Table). Of the 101 pts with an AFP test before start of second line therapy, 32.7% had AFP ≥400 ng/mL. Relationship between change in AFP and OS.Table: 708P
| Observed AFP change during 1L | N | Median OS (95% CI), months | Cross ratio | Kendalls Tau |
|---|---|---|---|---|
| Decrease (↓) to < 400 from ≥ 400 ng/mL | 20 | 14.3 (4.8-30.9) | 1 | 0 |
| ↓ ≥ 20 ng/mL | 85 | 7.4 (5.7-11.9) | 1 | 0 |
| ↓ ≥ 7 ng/mL/month | 68 | 6.8 (5.0-12.7) | 1 | 0 |
| ↓ ≥ 20% | 98 | 11.1 (7.3-13.2) | 1 | 0 |
| ↓ ≥ 50% | 51 | 12.2 (7.3-18.2) | 1 | 0 |
| Increase (↑) to ≥ 400 from < 400 ng/mL | 18 | 5.9 (3.6-7.4) | 1.506 | 0.202 |
| ↑ ≥ 20 ng/mL | 140 | 4.8 (3.7-5.5) | 1.902 | 0.311 |
| ↑ ≥ 7 ng/mL/month | 124 | 4.5 (3.1-5.2) | 2.006 | 0.335 |
| ↑ ≥ 20% | 141 | 5.2 (4.1-6.5) | 1.841 | 0.296 |
| ↑ ≥ 50% | 109 | 5.4 (4.5-6.9) | 1.755 | 0.274 |
Conclusions
Increases and decreases in AFP during 1L, regardless of AFP change definition, were generally associated with shorter and longer OS, respectively. Conclusions are limited by the risk of immortal time and selection bias, as not all patients had multiple AFP measures recorded.
Clinical trial identification
Legal entity responsible for the study
Eli Lilly and Company.
Funding
Has not received any funding.
Editorial Acknowledgement
Disclosure
L.M. Hess, Z.L. Cui, A. Girvan, P.B. Abada: Employee: Eli Lilly and Company. All other authors have declared no conflicts of interest.