Purpose To compare graft survival and visual outcomes for endothelial keratoplasty (EK) after a first penetrating keratoplasty (PK), with outcomes of repeat PK after a first PK.

Methods 400 eyes with a second graft (65 EKs, 335 PKs) performed after failure of a primary PK were identified through the Australian Corneal Graft Registry, a national prospectively followed cohort. Grafts were performed after January 2008 (follow-up of the second graft extending to 6.75 years maximum). Kaplan–Meier graft survival plots were constructed and Cox proportional hazards regression was used to identify independent risk factors for graft failure. Best-corrected Snellen visual acuity (BCVA) at last follow-up was compared with pregraft acuity.

Results Poor Kaplan–Meier graft survival was observed for PK-EK compared with PK-PK (log-rank=29.66, p<0.001). Variables retained in multivariate analysis as significantly influencing survival of the second graft included graft type (PK-EK or PK-PK, p<0.001), length of survival of the previous PK (global p=0.011), graft era (global p=0.018), occurrence of rejection in the second graft (p=0.005) and a history of raised intraocular pressure at any time (p=0.048), but not indication for the first graft. BCVA improved in the majority of surviving grafts and attainment of 6/12 vision was similar for both PK-EK and PK-PK groups.

Conclusions Our registry findings suggest that repeat PK may deliver a better outcome in terms of graft survival than EK after a failed PK that was performed initially for keratoconus or pseudophakic bullous keratopathy. For surviving grafts, visual outcomes appeared equivalent across groups.