Abstract
Purpose :
To evaluate the impact of automatically segmented intraretinal fluid (IRF) volume measurement on retreatment decisions (RETD) in comparison to the performance during DRCR Retina Network (DRCR) Protocol T based on CST/BCVA.
Methods :
During the study 656 patients/8160 visits have received 6048 treatments (INJ). In short, patients received INJ when CST/BCVA changed ≥10%/5 letters; otherwise, stability (STAB) was considered and INJ was paused after 2 STAB visits after W24 or CST<250µm/BCVA≥20/20. For this analysis, IRF volumes of STAB visits were analyzed by a deep learning algorithm (RetInSight Fluid Monitor Version 2, Vienna, Austria). IRF fluctuations (FLUC) ≥30nl were considered instable, whereas IRF STAB is defined as FLUC≤10nl. The positive predictive value (PPV) of CST STAB for IRF STAB is calculated, excluding 10-30nl FLUCs. It is analyzed in how many cases when CST STAB, but IRF FLUC were detected, the next visit received INJ due to CST/BCVA change. Analyses, content and conclusions presented herein are solely the responsibility of the authors while the data source was shared by the DRCR.
Results :
1192 loading INJ and 2437 INJ due to BCVA/CST change have been given. The remaining 2238 INJ were given under CST/BCVA STAB conditions. Only 713 (31.9%) showed IRF STAB as well (mean FLUC of 4nl), of which 493 (69.1%) remained STAB for the next visit (correctly classified as STAB, PPV of CST/BCVA 28.7%). Out of 2238 INJ, 1003 (44.8%) OCT showed an IRF FLUC≥30nl (mean change 144nl). 671 (66.9%) show a FLUC≥30nl again at the next timepoint, which in 357 (53.2%) cases now led to a relevant CST (228)/BCVA (192) change triggering a per-protocol INJ due to "instability". Considering all STAB timepoints (4026) (with/without INJ), the ones that received INJ at the next visit (2229) had a mean prior IRF FLUC of 69nl (mean CST change -2µm), whereas the ones that had received no INJ (1797) at the next visit, had a mean prior IRF FLUC of 26nl (mean CST change ±0µm)(p<0.0001).
Conclusions :
Automatically quantified IRF for treatment guidance instead of BCVA/CST can detect anatomic instability earlier and can prevent inadequate pausing INJ in case of fluctuations not mirrored in CST. In clinical routine visual outcomes may be improved by avoiding vision loss due to untreated IRF using objective fluid monitoring tools.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.