Eighty-eight eyes from 88 patients scheduled for routine cataract surgery were selected prospectively. Baseline patient data are shown in Table 1. Exclusion criteria for the study were eyes with significant corneal opacity, previous intraocular surgery, trauma, glaucoma, uveitis, Fuchs' endothelial dystrophy, other abnormalities that could cause significant endothelial cell impairment independent of surgery, eyes with small pupils that required iris retractors, and eyes with intraoperative complications such as posterior capsule rupture or postoperative complications.

Baseline patient data are shown in Table 1. Changes of CD, CV and HA over time are shown in Table 2. Our data shows the wound-healing process of corneal endothelium after cataract surgery, and we observed CV recovery at two months postoperatively.

Mean Densitometry_pentacam values according to grading by LOCS III classification are shown in Table 3. No eyes in our study had a nuclear opacity grade of 6 by LOCS III classification. A significant degree of concordance was found between Densitometry_pentacam and LOCS III classification (τ_b = 0.414, p = 0.000) for nuclear opacity.

Table 4 summarizes the univariate associations of the seven variables with the percentage of endothelial cell change (CD, CV and HA).

Univariate analysis demonstrated that ACV and Densitometry_pentacam were associated with the percentage decrease of CD at postoperative day 3. ACD_pentacam, ACV, amount of viscoelastics, and Densitometry_pentacam were associated with the percentage decrease of CD two months postoperatively. ACD_pentacam was associated with the percentage change of CV at two months postoperatively.

A multiple linear regression analysis was performed to identify the best set of independent predictors for the percentage decrease of CD. These data are shown in Table 5. The final model identified Densitometry_pentacam, viscoelastics and Pupil_CCC as independent predictors of decreased CD at postoperative day 3 (adjusted R² = 31.3%) and Densitometry_pentacam, viscoelastics and ACV as independent predictors of decreased CD at two months postoperatively (adjusted R² = 29.2%).

We found that intraoperative Pupil_CCC increased after injection of viscoelastics. However, we cannot compare these two values directly because Pupil_CCC was measured intraoperatively at the papillary plane by an injection needle of known length, and Pupil_preop was measured using the slit-lamp biomicroscopy scale.

Fig. 2 shows the correlation plot between ACD_pentacam and ACD_sono (r = 0.823. p = 0.000). Fig. 3 shows the correlation between ACD_pentacam and ACV (r = 0.650, p = 0.000). Fig. 4 shows the correlation between ACV and amount of viscoelastic (viscoelastic_amount; r = -0.021, p = 0.85). Fig. 5 shows the correlation between ACD_pentacam and viscoelastic_amount (r = 0.123, p = 0.253).

The Pentacam allows evaluation of the entire anterior segment from the anterior corneal surface to the posterior lens surface using a rotating Scheimpflug camera [1-7]. The noncontact measuring process takes two seconds and performs 12 to 50 single captures. The Pentacam has various functions, including the ability to measure ACV and nuclear density, which can be expressed as continuous numeric data. Until now, we could only evaluate nuclear opacity by categorizing with systems such as LOCS classification [7-9]. The Scheimpflug image provides the basis for objective, precise quantification of lens opacity [7]. The lens density is standardized from 0 to 100. Therefore, 0 means the lens shows no clouding, and 100 means the lens is completely opaque (Pentacam instruction manual, Oculus Inc.). Although we observed lens opacities from of LOCS III classification 1 to 5 in our study, we only obtained a relatively small range of Densitometry_pentacam values from 7.10 to 34.10%. We might have subjectively overestimated the lens opacity using the LOCS III classification.

Since Densitometry_pentacam data are not categorical data like LOCS classifications, we could more easily access statistical analysis using the former data type. We found a significant concordance between the LOCS III and Densitometry_pentacam methods.

Anterior chamber values are necessary for calculating intraocular lens power when planning surgical procedures. Precise anterior chamber measurements are essential for performing exact biometry and for surgical planning [5,6].

Several methods for measuring the anterior chamber depth are available [5,6,10]. The Pentacam and standard ultrasound devices define ACD as the distance between the anterior surface of the cornea and the anterior surface of the lens.

Ultrasound devices have become the most commonly used methods to measure ACD. Contact devices such as the ultrasound have disadvantages, including corneal indentations, the risk for corneal abrasions and infections, off-axis measurement, and possible precorneal echo spikes [10]. Pentacam imaging is a relatively new, noncontact automatic optical technique in ophthalmology. Although different methods to determine ACD may yield significantly different results [3,10], mean ACD measurements of the Pentacam and ultrasound were significantly correlated in our study (r = 0.823, p = 0.000), a finding that was similar to previous reports [2,6].

The Pentacam can evaluate ACV as well as ACD [4,5]. Both our study and previous reports found a good correlation between ACD and ACV (r = 0.650) [1,5].

We tried to evaluate LT using the Pentacam (LT_pentacam) and to compare these values with those obtained by ultrasound (LT_sono) in 50 eyes. There was a significant difference between LT measured by Pentacam (mean ± SC, 2.13 ± 0.62) and by ultrasound (mean ± SD, 4.42 ± 0.79) (paired t-test, p<0.05). The Pentacam LT values in our study were lower than other reports [11,12] and it was impossible to constantly measure LT_pentacam correctly. We suspect that this is because the posterior surface of the lens cannot be precisely determined with Scheimpflug imaging, which depends on pupil dilation. Previous reports have emphasized pupil dilation for measuring LT_pentacam (Pentacam instruction manual). Based on this information, we used the value of LT_sono to evaluate the effect on postoperative endothelial cell loss.

In our study, LT_sono showed a distribution of 2.82 to 5.50 with a mean value of 4.26 ± 0.66 mm, which is comparable to findings of other reports [11,12]. Jivrajka et al. [12] reported that LT tended to be thicker in older patients and in shorter eyes. In our study, LT_sono showed no correlation with axial length (r = -0.035, p = 0.748) and had only a weak correlation with age (r = 0.203, p = 0.058). We found that LT_pentacam was unsuitable for determining true lens thickness.

Several studies that evaluated the percentage of endothelial cell loss after phacoemulsification have been published [13-15]. The average loss reported after phacoemulsification varies between 4% and 25% [13]. In our study, the percentage loss of endothelial cell density at two months postoperatively was about 16%. Operative factors possibly associated with corneal endothelial injury include ultrasound energy, turbulence of the irrigating solution, instrument contact, air bubbles, touch by nucleus fragment, axial length, phacoemulsification time, nucleus grade, anterior chamber depth, and age [13-15]. Recent advances in endocapsular phacoemulsification procedures, instruments and viscoelastic substances appear to have helped decrease the degree of endothelial damage. At first, we intended to evaluate the relationship between seven anterior segment parameters and three endothelial cell indices (CD, CV and HA), but we did not observe a correlation between either anterior segment parameters and CV or anterior segment parameters and HA, except for the relationship between ACD and CV at two months postoperatively (Table 4). Therefore, we evaluated the correlation between anterior segment parameters and CD alone in our final regression model (Table 5).

We expected that the amount of injected viscoelastics required for adequate lens-iris diaphragm backward movement (and thus for safe capsulorrhexis) would be correlated with ACV or ACD_pentacam. However, there was no correlation between ACV and viscoelastic_amount (r = -0.021, p = 0.85) or between ACD_pentacam and viscoelastic_amount (r = 0.123, p = 0.253). We suspected the reason is that viscoelastic_amount use involves pushing back the lens-iris diaphragm for adequate capsulotomy, which is also affected by zonular laxity and posterior (vitreous) pressure of each patient during surgery, not only by the real total ACV. In our study, the amount of viscoelastics used was a factor affecting endothelial cell density at both three days and two months postoperatively. Additionally, small pupils are an added risk with any technique for cataract extraction [16,17].

Although the endocapsular technique remains popular, anterior chamber phacoemulsification may be more advantageous in specific cases, such as those with lens subluxation, unstable zonules or extremely hard nuclei. In these cases, there is less risk of injury to the posterior capsule if the procedure is performed in the anterior chamber [18]. Also, in small-pupil phacoemulsification, a method can be used that manipulates part of the lens into the anterior chamber through the pupil to maintain the pupil in a semidilated state [19]. Therefore, in small pupil phacoemulsification, pupil size can be a factor that affects corneal endothelial cells.

Although we cannot compare Pupil_CCC and Pupil_preop directly due to different measuring scales, our results showed that Pupil_CCC (rather than Pupil_preop) could have an effect on postoperative endothelial cell density.

We concluded that preoperative values measured by Pentacam were suitable for measuring all anterior segment parameters except LT, and anterior segment values measured by Pentacam were good predictors for postoperative endothelial cell loss.