Preop variables poorly predict graft dislocation after DSAEK

Study results show a strong correlation between the number of dislocations and rate of endothelial cell loss.

FORT LAUDERDALE, Fla. — Factors that were supposed to predict the chance of donor disc dislocation after Descemet’s stripping automated endothelial keratoplasty proved unreliable in foretelling dislocations, according to a study presented here.

Graft success was accurately predicted in only 8% of cases, despite robust clinical and statistical analyses, George O. Rosenwasser, MD, said at the Association for Research in Vision and Ophthalmology meeting.

“The predictive variables we studied were not useful in determining dislocation events, and the other variables should be examined to try and find out what is causing these,” Dr. Rosenwasser said.

The potentially predictive variables included patient age, concurrent phacoemulsification, graft size, incision size, donor death to surgery time (in days), donor endothelial cell density, donor age, diagnosis for transplant, presence of an anterior chamber or posterior chamber IOL, previous glaucoma surgery, and concurrent vitrectomy.

In addition, graft dislocation and rate of endothelial cell loss were found to be significantly related.

The prospective study included 412 eyes of 384 patients who underwent DSAEK performed with the donor shovel insertion technique, which involves minimal tissue handling and endothelial cell loss, Dr. Rosenwasser said.

Patients were assessed preoperatively and at 6, 12, 18, 24 and 36 months postoperatively.

The most common indications were Fuchs’ dystrophy and pseudophakic bullous keratopathy; 281 cases (68%) had Fuchs’ dystrophy. Indications also included primary graft failure.

Procedures and analysis

The study group included 204 patients with posterior chamber IOLs and 27 with anterior chamber IOLs.

Dr. Rosenwasser said he converted the remaining phakic cases to triple procedures: DSAEK with concurrent phacoemulsification.

“I convert patients with mild to moderate lens changes to a triple,” Dr. Rosenwasser said. “I prefer this method because it protects the donor tissue from having a cataract procedure later. The likelihood of morbidity to the donor and possible graft failure is high even in uncomplicated cataract surgery. Given the age of the population requiring these grafts and my preference for lifelong steroids at daily dosing, cataract acceleration or development is nearly unavoidable. In addition, a triple puts the patient in the OR only once and exposes them to that risk only once.”

DSAEK under penetrating keratoplasty was performed in 20 eyes.

Dr. Rosenwasser performed all procedures using a 5- to 6-mm incision with a 60/40 fold of an 8- to 8.75-mm donor lenticule. He used a running three-bite 10-0 nylon suture to close the wound, thoroughly removed viscosurgical material, filled the eye with air for 10 minutes and left an air bubble about the size of the donor lenticule.

“I close the wound and then basically start an infusion line and unfold the donor with an air injection if it does not unfold spontaneously,” Dr. Rosenwasser said. “Gravity infusion line [as a] first step frequently will allow me to unfold the graft spontaneously. I put in a full air bubble, squeegee it a little bit and then basically remove some of the air at the end, leaving a bubble the size of the donor disc.”

Dr. Rosenwasser and colleagues performed statistical analysis using logistic regression to arrive at a binary, or “on/off,” outcome: dislocation or nondislocation.

“The analysis shows how we formed a predictive model,” Dr. Rosenwasser said. “It tries to tell us if we got an adequate predictive model or not.”

In addition, the authors performed a multivariate analysis of variance to assess endothelial cell loss at 6 months, 12 months and 24 months in patients with no dislocations, one dislocation or multiple dislocations. Student’s t-test was used to compare preoperative and postoperative values.

Dislocation and endothelial cell loss

Study data showed that single re-bubbling was required in 21% of eyes, double re-bubbling was required in 4% of eyes and triple re-bubbling was required in 2%. Three patients required a fourth re-bubbling and one was converted to PK. Primary graft failures occurred in four eyes (less than 1%), Dr. Rosenwasser said.

The prediction model for graft dislocation proved statistically insignificant, accurately predicting dislocation in only seven eyes (8%), Dr. Rosenwasser said.

“We had no single variable that could predict [dislocation] or that had a significant odds ratio change that would help us to figure out what’s making these [grafts] come off,” Dr. Rosenwasser said. “It lacked predictive significance with a poor classification, a 92% miss rate. If you miss 92% of the time, you’d think, ‘Well, something’s not going too well here.’”

Data showed that at 1 year postop, about 84% of patients had visual acuity of 20/40 or better, and 24% had visual acuity of 20/20 or better, Dr. Rosenwasser said.

Results also showed that the endothelial cell loss rate was 31% among patients with no dislocations and 47% among patients with multiple dislocations at 1 year.

“That’s the only thing that’s statistically significant in terms of losing cells,” Dr. Rosenwasser said. “The others are all blissfully random. … The bottom line is, if you have multiple dislocations, you lose more cells.”

The between-group differences in endothelial cell loss at 1 year were statistically significant (P = .047).

Hypotony stemming from previous glaucoma surgery, patient noncompliance with supine positioning and donor tissue quality may contribute to donor disc dislocation, Dr. Rosenwasser said. – by Matt Hasson

• George O. Rosenwasser, MD, can be reached at Central Pennsylvania Eye Institute, 825 Fishburn Road, Hershey, PA 17033; 717-533-5200; fax: 717-533-2606; e-mail: gr@cpeye.com.

This ARVO presentation by Rosenwasser, Lu and Friend did a statistical analysis with logistic regression techniques to determine if there was any factor of either donor characteristic or patient characteristic to predict the event of donor tissue dislocation in DSAEK surgery. The series was large (412 eyes) done by a single surgeon. Although their dislocation rate was relatively high (21%), their analysis revealed no predictive factors. Their findings confirmed those of our prior publications (Chen et al 2008, Terry et al 2009) which concluded that careful surgical technique is the single most important factor in preventing dislocation. Removal of interface fluid, peripheral recipient bed scraping and minimizing trauma to the donor endothelium through each stage of the procedure can yield a less than 2% rate of dislocation, regardless of donor characteristics or clinical setting.

The second aspect of the authors’ paper confirmed once again that donor dislocation is not an innocuous event, but results in a significantly higher cell loss and likely shorter graft survival.

– Mark A. Terry, MD
Director, Corneal Services, Devers Eye Institute, Portland, Ore.