The results of a corneal transplant, like that of any ocular surgery technique, can be considered from two main points of view: anatomical and functional. Given that corneal transparency is perhaps the property most directly related to the visual function, it largely summarizes both aspects. The role of the cornea in the optics of the eye makes the refractive results have a great functional importance, especially astigmatism. This is, however, a facet on which there is abundant information in the literature^1,2, we have published our experience^3,4, and exceeds the scope of this work. Therefore, we will concentrate on the survival of the grafts and the prognostic factors that influence it.

THE ANALYSIS THROUGH ACTUARIAL LIFE TABLES AND SURVIVAL CURVES

The results of the corneal grafts present two characteristics that condition their study. On the one hand, failure is in general an irreversible state, which naturally coincides with a survival model. On the other hand, it is an issue that requires a long-term study. Since keratoplasty is not a particularly frequent type of surgery, obtaining prospective series of adequate size and follow-up time is not an easy task. For this reason, as in many similar situations in medicine, analysis by life tables and actuarial survival curves (or Kaplan-Meier) is very useful. These have the virtue of partially compensating the incomplete information – because it is retrospective and with different follow-up times – since they give each case an influence on the calculation of the curve only during the time it has been followed.

It must be remembered that the strength of the actuarial method is supported by its origin in the estimation of risks by insurance companies, which could never be allowed to operate based on unreliable calculations. And although possible inclusion bias or other inhomogeneity cannot be excluded (in the technique, post-operative treatment, etc.), these are limited when dealing with series of a single center.

For those not very familiar with this methodology, a brief reminder. Survival curves are based on a very simple but iterative calculation. It starts from 100% at the time of the intervention (time «zero» for each case) and from there the curves go down. The event is defined (here, the graft failure) and for each period of time (one month, one trimester, etc.) the number of patients that are controlled and the events that have occurred are accounted for. The risk of the event is obtained and subtracted from 1 (100%), which gives us the survival value for that period. In each successive period, the impact of the events is counted on the total of cases "at risk", that is, those that are transparent and under control. Although not all patients are controlled in all periods, if the transparency is confirmed even at the end of years, we can infer that it has been like this during all that time.

When interpreting the curves, keep in mind that: 1) The percentages at a given time are not real survival data but only estimated on a partial sample – since 100% of the series is no longer controlled; 2) The validity of each point of the curve depends on the number of cases followed at that time, which is decreasing: it is not correct to judge them by the final percentage, since it usually represents a small group of cases, where a single failure makes the % fall sharply. When comparing the curves, this should be done as a whole. To do this, tests like the one of logrank are used, which is nothing more than a chi-square calculated cumulatively, comparing the percentages for each monitoring period.

When it comes to values ​​that can be stratified numerically, such as grafting order, diameters, ages, times, etc., a trend test can also be done, which consists of assigning an ordinal factor to each category (1,2,3, etc.), which is incorporated into the calculation of logrank. The result is the possible detection of trends, even where the differences are not significant or, on the contrary, there may be statistically significant differences but, if they lack a tendency, they may not have any real meaning. We'll see some examples later.

RESULTS OF THE CLASSICAL SERIES

In the literature there are numerous reviews of series of penetrating keratoplasty (PK) followed in the long term^5-8. These studies show that the most important prognostic factors are diagnosis, corneal vascularization of the recipient and diameter of the graft. In 1988 we conducted a study of 2,886 PKs performed successively at the Centro de Oftalmología Barraquer (Barcelona) between 1956 and 1987⁹. The follow-up was up to 31 years, with a minimum of 3 months. A simple accounting of the number of observed failures indicates that most of them occur in the 1^st year and cumulatively 27% of the total fail (Figure 1).

Influence of Graft Order No.

When applying the survival curves, the first factor analyzed was the No. of order of the grafts (1^st = 88.6%; 2^nd = 8.8%; ≥ 3^rd = 2.5%), which show a 5-year survival rate of 75%, 30% and 10% respectively, extremely significant both, the difference and the tendency (p <0.00000001) (Figure 2).

Influence of the etiology or indication

We divided the causes of the transplant into 12 categories, as explained in the chapter on indications (3.1.1). Keratoconus – apart from being the most frequent – was the most successful at 5 years (90%) followed by stromal dystrophies and inactive leukomas (70-80%), endothelial dystrophies and trachoma (60%), trauma and others (55%), active keratitis and tectonic grafts (45%), aphakic and pseudophakic edema (30%), re-grafts (25%) and finally burns and chemical injuries (15%) (Figure 3).

The differences were as a whole, as expected, extremely significant, but interestingly, when analyzing each pair of diagnoses, the keratoconus group was already significantly better (p <0.01) than the next one of stromal dystrophies. The poor prognosis of aphakic and pseudophakic edema was significantly worse than the previous group of "in hot" PKs (p = 0.03) and no better than the next group of re-grafts (p = -0.66). This may reflect that in this period these patients (3.6% of the total) used to be intervened late and with other complications such as glaucoma, etc., which overshadowed the prognosis.

Influence of vascularization

The vascularization of the recipient cornea is one of the most recognized risk factors in PK. This series corroborates this, both in the global analysis and in each etiology separately – although the distribution of vascularity varied greatly according to diagnoses. We defined 3 groups as seen in the preoperative pictures: avascular (37%), mild-moderate or 1-3 quadrants (50%) and severe or 4 quadrants (13%). The 5-year survival was, respectively, around 90%, 60% and 20%, with extremely significant differences (Figure 4).

The only group in which this factor was not significant (p = 0.23) was that of PK "in hot", where only 2 cases of 169 were avascular. In keratoconus, the situation was reversed, with only a minority of cases with mild-moderate (1.6%) or severe (0.3%) vascularization. Here the difference did not reach significance, but the trend did (p = 0.02).

Influence of graft diameter

The diameter of the graft is another of the most known risk factors, since the proximity to the limbus would reduce the immunological privilege of the cornea. It is, however, an influence not as clear as vascularization, and not all studies coincide¹⁰. In our series we observed that the worst results were with the largest diameter, but also with the smallest. Thus, 5-year survival rate was 70% for the two groups 7.5-7.9 and 8.0-8.4 mm; 50% for the 7.0-7.4 mm; 40% for the 8.5-9.0 mm; 35% for the <7.0 mm, and <10% for those >9.0 mm. The set of curves adopts a "saddle" shape (Figure 5), which corresponds to a significant trend test (p = 0.002) but much less than the logrank difference achieved (p <0.0000001).

In some group, as in the re-grafts, the difference was significant (p <0.0001) but the trend was not (p = 0.29). These findings can be interpreted as a bias of the smaller diameters towards cases of worse prognosis due to the etiology or the corneal state. If we compare the general distribution of diameters with that of the different diagnoses, it is quite variable. In keratoconus, grafts of intermediate diameters (7.5-8.4 mm), more favorable, were used in 92.6%; on the other hand, there is a more wide-spread distribution in the re-grafts, with 27.6% of small diameters (≤ 7.4 mm) and 13.1% very large (≥ 8.5 mm) (Figure 5, tarts).

Influence of the patient's age and sex

The age of the patient is an important factor, although probably linked to the etiologies. In general, we find the best results in the group of young people between 11 and 30 years old, with 5-year survivors of 80%, which after that age are going down: 65% in the age of 31-50 years, 50% in the 51-70 years and 40% in the ≥ 71 years, while it is 60% in the pediatric group up to 10 years of age (Figure 6).

Probably the best results of the young people are due to the predominance in that group of keratoconus (63.9%), while 76% of aphakic/pseudophakic edemas, with markedly unfavorable results in this series, were ≥ 51 years old. The age distribution is more similar to the general distribution in other groups, such as stromal dystrophies or re-grafts (Figure 6, tarts). The statistical significance was again very high for the global series (p <0.0000001), as well as for almost all diagnostic groups separately (although with more discrete p-values).

In the case of keratoconus and re-grafts, age did not reach logrank significance (p = 0.09 and p = 0.07, respectively) but the trend did (p = 0.03 and p = 0.04). It is striking that this tendency was to worsen with age in cases of keratoconus – in the general line of the ages – but it tended to improve among the re-grafts. This could be explained because the cases of keratoconus would be more advanced with age, while in the re-grafts older ages would be associated with a lower aggressiveness of the immune system. The sex of the patient did not show significant influence (p = 0.27).

Donor factors

There were no significant differences by age (p = -0.57) or donor sex (p = 0.21). It should be noted that 20.5% of the donors were over 80 years old. In the three-dimensional graph, it gives the impression that the survival curves are getting worse, but the trend test also did not reach the significance (although it was closer, with p = 0.059). We should recall that the last histogram always has fewer cases and you have to compare the curves as a whole. Even when separating in two groups of more or less than 40 years, significance was found (p = 0.11) (Figure 7). When comparing the age distributions of the donor in each diagnostic group, no differences were observed.

With regard to gender, the curve was somewhat better for female donors, without being significant in the overall series (p = 0.21), although it did so in the age groups of young patients (11-30 years, p = 0.04) and the trachoma group (p = 0.05). The meaning of this is unknown. Neither did the coincidence or not of sexes between patient and donor have significance.

Conservation factors

In this series, all the globes were kept whole in fresh in a humid chamber at 4 °C, without the use of culture media. The death to enucleation time in general, was attempted not to exceed 8 hours (h), which occurred in 4.3%. Survival according to this time presented significant logrank differences (p = 0.015), but the trend did not (p = -0.59). This indicates that there is no temporal progression, so these differences do not have a clear meaning; as seen in the graph (Figure 8, above), the worst results were given with the group enucleated between 6-8 h, but also with that of only 1-2 h. Not even comparing more or less than 6 h is significant (p = 0.057).

With the death-graft times, which exceeded 24 h in only 3.4%, there were no significant differences in survival when comparing 6 groups (p = 0.19) and the trend only touched the significance (p = 0.059), but a p = 0.017 was obtained when comparing only 2 groups of less than or more than 12 h.

RESULTS OF A MOST RECENT SERIES

We have carried out a similar study, with a series of 966 consecutive PKs between the years 2001 and 2006. As far as possible, the same classification criteria of the classic series were followed, and the results are generally coincident, with some differences. The distribution of indications of this study is discussed in chapter 3.1.2; somewhat surprisingly, in this period the percentage of keratoconus still increased, up to 38%. Due to the limited space, the survival results and prognostic factors of this series will be available in a version of this chapter in the Spanish Ophthalmological Society (www.oftalmoseo.com).