Since Taylor1 described the triple procedure in 1976, many surgeons have adopted this technique to manage patients with reduced visual acuity (VA) as a result of corneal pathology in the presence of a cataract. The triple procedure became popular because it offers the potential of quicker visual rehabilitation than sequential surgery. Difficulties in predicting postkeratoplasty corneal power, however, soon became apparent.1 Today, there remains no precise and reliable method to predict the intraocular lens power that would produce the intended postoperative target refraction after the triple procedure. Clearly, although endothelial keratoplasty supersedes this problem, not all cases are suitable for this procedure and penetrating keratoplasty (PK) still has an important role. Sequential surgery, allows measurement of postkeratoplasty keratometry or topography for use in the calculation of the power of the intraocular lens as well as the modification of the postkeratoplasty refractive error at the time of cataract surgery.2
Although there have been studies supporting or advocating either triple procedure3–6 or sequential surgery,2,7,8 it remains unclear which type of surgery optimizes the refractive and visual outcome. In particular, many of these studies2,3–7 have relied on the use of spherical equivalents to assess outcome, which is relatively insensitive and may underestimate or not detect differences in refractive outcome between procedures.9 A typical example is the inability to distinguish between two refractive powers with similar spherical equivalents, e.g., −3 DS and −6/+6 × 90. Similarly, the errors introduced treating the cylinder and spherical components of a refractive power as independent variables has been well demonstrated by Harris.10,11 In addition, it has recently been shown that the spherical equivalent introduces a systematic bias in aggregate data or data containing high refractive powers and may lead to false statistical conclusions.12,13 The aim of this study was therefore, to compare the outcome of patients undergoing sequential and triple procedures including an analysis of the refractive outcome both in terms of the difference in refractive error and the variability of the refractive error without relying on the relatively insensitive and systemically biased spherical equivalent for this type of analysis.
The medical records of all patients who had undergone either a triple procedure or sequential surgery at St Paul's Eye Unit over a 10-year period were examined. This study has been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. All patients gave their informed consent for inclusion in this study and for publication. Patients were excluded if there was less than 1 year of follow-up from the last date of surgery. Surgery had been performed by two experienced corneal surgeons (S.B.K. and M.B.). A decision on the type of surgery, that is, sequential or triple was made by the surgeons according to the following criteria: a triple procedure had been performed when both corneal pathology and cataract were thought to be the cause of symptomatic reduced VA and the patient had not wished to undergo sequential surgery. A PK alone had initially been performed if corneal pathology had been thought to be the predominant cause of symptomatic reduced VA or if cataract had been present but the patient opted to have sequential surgery. Cataract surgery was then performed 6 to 9 months after PK. Only the first eye that had undergone surgery in each patient was analyzed.
Data recorded included the indication for surgery, age at the time of PK, sex, duration of follow-up, intra- and postoperative complications, preoperative and final best-corrected VA (BCVA), postoperative refraction and additional surgery performed. Preoperative axial length had been measured by A-scan ultrasonography (Model Axis, Class 1-Type B. Quantel Medical) and a Javal-Schiotz keratometer or Orbscan topography for simulated keratometry readings. The SRK II formula had been used for intraocular lens power calculation. For PK, the donor-host disparity was 0.5 mm. For the triple procedure, a continuous curvilinear capsulorhexis had been performed, the nucleus irrigated and expressed, or divided and aspirated following open sky phacoemulsification and the intraocular lens placed in the capsular bag. Donor corneas had been transferred without manipulation, using the method of Kaye et al.,14 and sutured using a combination of continuous and interrupted sutures.
For sequential surgery, small incision phacoemulsification had been performed at least 6 to 9 months after PK. The corneal section had been placed in the steepest meridian and additional relaxing arcuate keratotomies made as indicated. All patients had received postoperative prednisolone acetate 1% in addition to antimicrobials and other topical medications (such as for glaucoma) as appropriate. Postoperatively, sutures were removed to reduce astigmatism. Refractive and visual outcomes were used for analysis when refraction and corneal topography had been stable for at least 3 months after suture or surgical manipulation. Graft survival was measured using Kaplan-Meier survival curves.
Analysis of Refractive Data
The principal outcome measure used for comparing the two groups was the difference and variability between the actual and planned (target) postoperative refractive data. In addition, the groups were compared in terms of pre- and postoperative refractive error and variance thereof. Refractive data were transformed into Long's matrix formalism,15 i.e., f11 = S + C sin2a, f12 =−C sin a cos a, and f22 = S + C cos2a, where a, is the axis, and then analyzed according to the methods of Kaye and Harris.9–11 The data was then transformed back into standard spherocylinder notation using Keating's method15 except for the standard deviation because it is not possible to express the standard deviation of the data in standard notation. The data is therefore presented as mean refractive data, in standard notation, (S/C axis) and standard deviation as a row of the matrix, [ f11, f12, f22 ] in both the text and table.
For an analysis of the variability of the refractive data, the multivariate normal model was used to compare the two variance-covariance matrices using Bartlett's test statistic16 (see Appendix, Supplemental Digital Content 1, http://links.lww.com/OPX/A25), i.e.,
, where n is the number of observations in each group, 1 and 2 are sample covariance matrices for each group and is the common covariance matrix obtained from the formula
is the determinant of . This M statistic has a χ2 distribution on p(p + 1)/2° of freedom, where p is the number of variables in the multivariate observation, which is three for analyzing refractive data giving 6° of freedom. Refractive data were then transformed back into spherocylinder notation using Keating's method.15 In addition, to allow comparison with previous studies, the refractive data were also analyzed using final spherical equivalents or, more correctly, nearest equivalent sphere (NES) and target refraction. Tests for significant difference were based on the t-test, paired for within group and unpaired between groups.
Analysis of Visual Acuity
BCVA, by spectacles, was converted into logMAR units for calculation of the means of the pre- and postoperative visual acuities. A Student t-test was used to test for a difference between groups. In addition, the number of patients achieving a BCVA of 6/12 or better was compared in the two groups.
Overall, a total of 26 triple procedure cases were included. Two patients had had bilateral surgery so the second eyes were omitted. One patient had had a failed corneal transplant at 3 months after surgery. Twenty-three patients were therefore available for further analysis in the triple procedure group. Thirty cases were identified in the sequential surgery group. Four patients had Down's syndrome with mental impairment so that VA could not be accurately measured, one patient had been unsuccessfully treated for rejection 2 weeks after cataract surgery, one patient had developed primary graft failure and was re-grafted and one patient had had bilateral surgery. Twenty-three patients who had had sequential surgery were therefore also available for further analysis. Patients who had undergone the triple procedure were significantly older, p = 0.047 (average 71 years, range, 37 to 87 years) than those who had undergone sequential surgery (average 63 years, range 21 to 86 years). The female to male ratio was 16:7 in the triple procedure group compared with 7:16 in the sequential surgery group.
The indications for patients undergoing surgery in the triple and sequential groups are outlined in Table 1. Fuchs dystrophy was the most common indication in the triple procedure group and herpes simplex keratitis in the sequential surgery group. The mean follow-up after PK was 28.3 months (range, 1 to 6.9 years) in the triple procedure group and 49.7 months (range, 1.2 to 9.3 years) in the sequential surgery group. The mean follow-up after cataract surgery was 38.7 months (range, 1 to 6.25 years) in the sequential surgery group. Arcuate keratotomies had been performed in five patients from the triple procedure group and in seven from the sequential surgery group at the time of cataract surgery.
Sixteen patients in the triple procedure group had a final BCVA of 6/12 or better, whereas seven patients had a BCVA <6/12 because of amblyopia (3), age-related macular degeneration (2), congenital measles retinopathy (1), and unknown (1). Nineteen patients in the sequential surgery group had a final BCVA of 6/12 or better, whereas four patients had a BCVA <6/12 as a result of amblyopia (1), macular scarring after previous retinal detachment repair (1) and glaucoma (2). The mean postoperative BCVA was 0.40 (SD = 0.39) logMAR units in the triple procedure group and 0.23 (SD = 0.25) logMAR units in the sequential surgery group (Table 2). The postoperative BCVA was significantly better than the preoperative BCVA in both groups (p < 0.01 in both groups). There was no difference in either the preoperative, postoperative or difference in the change between pre- and postoperative BCVA between groups (p = 0.54, p = 0.09, and p = 0.12, respectively).
After sequential surgery, there was no statistically significant difference between the mean postoperative refractive outcome (−1.33/+0.79 × 175 SD = 2.39/1.16/2.33) and the mean target refraction (−0.43 DS, SD = 0.99) p = 0.48. For the triple procedure, although there was a greater difference between the mean target (−0.77 DS, SD = 1.66) and mean postoperative refraction (−2.73/+1.05 × 109 SD = 4.30/1.42/4.43), the difference was not statistically significant (p = 0.39). Similarly, the difference between the postoperative refraction and the target refraction in the sequential surgery group (−0.9/+0.79 × 175 SD = 2.21/1.16/1.91) was not significantly smaller (p = 0.35) or less variable (see Appendix, Supplemental Digital Content 1, http://links.lww.com/OPX/A25) than the difference in the triple procedure group (−1.96/+1.05 × 95 SD = 4.24/1.42/4.48). There were no significant differences (p > 0.1) between the two groups in pre- and postoperative refractive data or in the difference between the pre- and postoperative refraction (Table 2). In terms of the variability of the difference between target and postoperative refractive outcome, (Table 2) there was no difference between the sequential and triple groups (p > 0.99) (see Appendix, http://links.lww.com/OPX/A25).
Comparative analysis using NES produced similar results. There was no significant difference between the mean postoperative NES in the triple (−2.21 SD = 4.05) and sequential (−0.94 SD = 2.04]) surgery group (p > 0.1) In addition, there was also no significant difference between mean target and mean postoperative NES for either triple (−1.44 SD = 4.05) or sequential (−0.51 SD = 1.70) surgery (p > 0.1 for both), Table 2.
Adverse Events and Reactions
There had been no intra-operative complications in either group. Corneal transplant rejection occurred in two cases from the triple procedure group; one at 1 month and the other at 22 month after surgery. In the sequential surgery group, five patients experienced rejection; one patient had two episodes (one before and the other 17 months after cataract surgery), the other four cases of rejection had occurred before cataract surgery. All cases of rejection in both groups were successfully treated with either topical steroids or a combination of topical steroids and systemic steroids as deemed necessary. The corneal transplant survival rate was 91% for both triple procedure and sequential surgery. Graft (endothelial) failure was recorded in two cases from group one, one at 14 months and the other at 18 months after surgery. There were also two cases of graft (endothelial) failure from the sequential surgery group, one at 20 months and the other at 24 months after cataract surgery.
Endothelial keratoplasty is becoming the predominant method for treating patients with endothelial failure (e.g., Fuchs dystrophy) and may be associated with a more predictable refractive outcome when combined with cataract extraction and lens implant.17–19 PK, however, still has a major role particularly for patients with corneal stromal disease. In this study, more than 70% of the subjects were not suitable for endothelial keratoplasty. Many patients so affected have or will develop cataract. Thus, there is a need for information comparing the outcome of the triple procedure to that of sequential surgery. In this study, we have compared the results between patients undergoing triple procedure and sequential surgery in terms of VA, refractive outcome, complications, and transplant survival.
Many previous studies have relied on the use of postoperative NES to assess the refractive outcome of surgery. The problem with using NES is that larger refractive errors, particularly astigmatic refractive errors, may have low NES. This may give the impression that patients are close to target refraction when in fact the spread of the refractive outcomes may be wide. This is particularly so following PK where high levels of astigmatism are common, for example, the spherical equivalent of −2.00/+4.00 × 90 is 0 diopter. In addition, the spherical equivalent introduces a systematic bias in aggregate data or data containing high refractive powers and may lead to false statistical conclusions.12,13 Furthermore, as has been previously demonstrated,11 the co-dependence and lack of invariance of the sphere and cylinder prevent any meaningful statistical analysis when treating each component separately. In view of these issues, we used the complete refractive data for analysis based on Long's formalism15 and the methods of Kaye and Harris,9 and then compared the variance of the refractive data using Bartlett's procedure.16
In this study, patients undergoing the triple procedure were older than those having sequential surgery (71 vs. 63 years, p = 0.047) and Fuchs dystrophy was the most common indication for surgery in this group. Older patients and those with Fuchs dystrophy are more likely to have advanced lens opacities,20,21 which is the clinical scenario that was advocated by Taylor1 in proposing the triple procedure. Since the introduction of the triple procedure various modifications have been introduced, in particular to cope with changes in postkeratoplasty keratometry. Despite this, there has not been universal acceptance because of continued unpredictability of refractive outcome and issues of safety. Detractors of the triple procedure claim that the intra-operative risks are higher compared with sequential surgery, particularly because the open-sky period is longer, during which time the eye is most susceptible to the development of an intra-operative complications.3,4,22,23 It is difficult to determine the exact rate of posterior capsule tear and vitreous loss associated with the triple procedure. Some studies report a posterior capsule tear rate as high as 15% and vitreous loss of 4 to 12%.4,24 Others have noted a posterior capsule rupture rate of <1%.24 In our study, although relatively small, we did not encounter any episodes of posterior capsule tear, vitreous loss or suprachoroidal hemorrhage in either group so in our hands both procedures were equally safe.
Sequential surgery, while having the potential benefit of measurable keratometry and adjustment of refractive error, involves two separate operations and further risk to the corneal graft. In addition, the combination of postoperative inflammation and topical steroids may accelerate cataract formation. In this setting, one can argue that the triple procedure is advantageous because it avoids a second surgical procedure and the accompanying anesthetic and surgical risks.3,20 However, we did not find any difference in graft survival (91%) or adverse reactions between the two groups.
In terms of VA, although 83% of patients who had sequential surgery had BCVA equal or better than 6/12, there was no statistically significant differences between the groups. Although the absence of a difference is likely to be due to the variability in BCVA and sample size, it would suggest that there may be a clinical benefit in terms of BCVA for sequential surgery. This trend is also apparent in other reports where it is reported that 66 to 86% of patients who had sequential surgery achieved 6/12 or better2,3,7 as opposed to 65 to 72% of patients who had the triple procedure.3–6 Although Snellen VA charts are useful in detecting substantial changes in VA, it is possible that a difference would have been evident using more selective VA charts.
One of the major disadvantages with the triple procedure is accuracy of the intraocular lens power calculation. The accuracy of conventional intraocular lens regression formula relies heavily on predictable keratometric readings. The unpredictability of postkeratoplasty keratometric values following the triple procedure should cause intraocular lens power calculation errors and unexpected refractive outcomes. Binder using standardized suturing techniques, intraocular lens design and personal constants in regression formula reported an improvement from 50 to 59% of patients achieving refractive errors within 2 diopters (NES) of emmetropia using the triple procedure technique.5,25 The refractive outcome after sequential surgery, however, seems to be more accurate. Geggel2 in 1990 reported the results of 22 patients who had intraocular lens implantation after combined PK and extracapsular cataract extraction. In his series, 95% of patients achieved refractive errors within 2 diopters of the desired result. These studies, however, used the NES, which as already discussed can be misleading. This is evident by our results. For example, if one considers the NES, this was within 2D of target refraction for both groups, but as is also apparent the variance of the mean refractive error (which includes the astigmatism) was very high. It would therefore, be incorrect and misleading to conclude that the refractive outcome was within an acceptable range of the target refraction.
Our study has several inherent biases, which include non-randomization of patients and retrospective collection of data. In addition, the groups were not matched for age, diagnosis or graft size, all of which may affect outcomes of PK and cataract surgery. However, this series represents a group of subjects that the corneal surgeon commonly encounters. The study is also limited by the sample size. Given the variability of both the refractive and visual outcome, it is estimated that a sample size of at least 182 (80% power) would be need to detect a difference in BCVA. To achieve this size, requires a multicenter study. Although desirable, this introduces further variables such as multiple surgeons and observer variability.
In summary, when planning surgery for patients with both corneal pathology and visually significant cataract, although the current evidence suggests that sequential surgery offers no significant advantage over the triple procedure in terms of refractive predictability or variability, there is a suggestion of a trend favoring sequential surgery both in terms of BCVA and number of patients achieving 6/12 or better vision. This, along with other factors such as the need for rapid visual rehabilitation, medical illnesses and anesthetic risk are important considerations to help select the most appropriate procedure for the patient.
The authors do not have any financial or proprietary interest in any material or method mentioned.
Countess of Chester NHS Trust Hospital
Health Park, Liverpool Road, Chester CH2 1UL
The appendix is available online at http://links.lww.com/OPX/A25.
1. Taylor DM. Keratoplasty and intraocular lenses. Ophthalmic Surg 1976;7:31–42.
2. Geggel HS. Intraocular lens implantation after penetrating keratoplasty. Improved unaided visual acuity, astigmatism, and safety in patients with combined corneal disease and cataract. Ophthalmology 1990;97:1460–7.
3. Pineros OE, Cohen EJ, Rapuano CJ, Laibson PR. Triple vs nonsimultaneous procedures in Fuchs' dystrophy and cataract. Arch Ophthalmol 1996;114:525–8.
4. Crawford GJ, Stulting RD, Waring GO III, Van Meter WS, Wilson LA. The triple procedure. Analysis of outcome, refraction, and intraocular lens power calculation. Ophthalmology 1986;93:817–24.
5. Binder PS. Intraocular lens powers used in the triple procedure. Effect on visual acuity and refractive error. Ophthalmology 1985;92:1561–6.
6. Hunkeler JD, Hyde LL. The triple procedure; combined penetrating keratoplasty, cataract extraction and lens implantation. J Am Intraocul Implant Soc 1979;5:222–4.
7. Hsiao CH, Chen JJ, Chen PY, Chen HS. Intraocular lens implantation after penetrating keratoplasty. Cornea 2001;20:580–5.
8. Hamill MB. Sequential surgery may be the best approach for the patient. Arch Ophthalmol 2000;118:415–7.
9. Kaye SB, Harris WF. Analyzing refractive data. J Cataract Refract Surg 2002;28:2109–16.
10. Harris WF. Testing hypotheses on dioptric power. Optom Vis Sci 1992;69:835–45.
11. Harris WF. Invariance of ophthalmic properties under spherocylindrical transposition. Optom Vis Sci 1997;74:459–62.
12. Kaye SB. Approximating lens power. Optom Vis Sci 2009;86:382–94.
13. Kaye SB. Average focal length and power of a section of any defined surface. J Cataract Refract Surg 2010;36:665–70.
14. Kaye SB, Lagnado R, Morton CE, Kaye DB. Atraumatic transfer of the donor corneal button. Cornea 1997;16:240–1.
15. Long WF. A matrix formalism for decentration problems. Am J Optom Physiol Opt 1976;53:27–33.
16. Krzanowski WJ. Principles of Multivariate Analysis: A User's Perspective. Oxford: Oxford University Press; 1990.
17. Keating MP. On the use of matrices for the mean value of refractive errors [letter]. Ophthalmic Physiol Opt 1983;3:201–3.
18. Covert DJ, Koenig SB. New triple procedure: Descemet's stripping and automated endothelial keratoplasty combined with phacoemulsification and intraocular lens implantation. Ophthalmology 2007;114:1272–7.
19. Terry MA, Ousley PJ. In pursuit of emmetropia: spherical equivalent refraction results with deep lamellar endothelial keratoplasty (DLEK). Cornea 2003;22:619–26.
20. Pineros O, Cohen EJ, Rapuano CJ, Laibson PR. Long-term results after penetrating keratoplasty for Fuchs' endothelial dystrophy. Arch Ophthalmol 1996;114:15–8.
21. Martin TP, Reed JW, Legault C, Oberfeld SM, Jacoby BG, Yu DD, Dickens A, Johnson HP. Cataract formation and cataract extraction after penetrating keratoplasty. Ophthalmology 1994;101:113–9.
22. Baca LS, Epstein RJ. Closed-chamber capsulorhexis for cataract extraction combined with penetrating keratoplasty. J Cataract Refract Surg 1998;24:581–4.
23. Price FW Jr, Whitson WE, Ahad KA, Tavakkoli H. Suprachoroidal hemorrhage in penetrating keratoplasty. Ophthalmic Surg 1994;25:521–5.
24. Davis EA, Azar DT, Jakobs FM, Stark WJ. Refractive and keratometric results after the triple procedure: experience with early and late suture removal. Ophthalmology 1998;105:624–30.
25. Binder PS. The triple procedure. Refractive results. 1985 update. Ophthalmology 1986;93:1482–8.