Patients and methods

Twenty nine patients (14 female, 15 male) with acquired anophthalmia were examined clinically and with high resolution computed tomography (CT). Patients with conditions possibly causing facial asymmetry, such as orbital fractures, phthisical eye, congenital or acquired microphthalmia, and buphthalmos before enucleation, were not included. Eight patients with long standing anophthalmia of 25–52 years had been enucleated in childhood aged between 0.4 and 8 years (group I); one half of them received a primary orbital implant (18 mm Allen implant). The remaining 21 patients were enucleated in adulthood aged between 15 and 53 years. Fifteen of these patients (group IIa) also had long standing anophthalmia of 7–53 years, whereas six patients (group IIb) had short follow ups of 9 months to 4 years after enucleation. Four patients before enucleation were used for controls.

A CT scanning technique with 1.5 mm contiguous slices in a transverse plane was applied with the patients in a supine position (Philips Tomoscan SR 7000, Eindhoven, Netherlands). Three dimensional reconstruction was performed based on interactive segmentation of the cross sectional CT images at a Gyroview HP workstation (Philips Medical Systems, Eindhoven, Netherlands) with Gyroview software, version 5.0.3 (ISG Technologies Inc, Mississauga, Canada). The lower threshold for orbital volume segmentation was −204 H, the upper +166 H. High density structures such as orbital implants or artificial eyes were segmented separately, using thresholds of +166 H and +2911 H (maximum). For low density structures such as free air behind an artificial eye the thresholds were −204 H and −1024 H (minimum) respectively. The margins of the bony orbit were defined by the orbital walls and the orbital entrance, which was formed by the superior, lateral and inferior orbital rim, the anterior lacrimal crest, and a line connecting the anterior lacrimal crest with the superior orbital rim.9 The optic canal, the nasolacrimal duct, and the pterygopalatine fossa were not included (Fig 1). The volume confined by these areas, completed by the volume of implants, artificial eyes, or air, if present, is defined as the “bony orbital volume” (BOV). The volumes were calculated by voxel counting. Analysis and evaluation in this study are based on intraindividual differences between BOV of the anophthalmic and the fellow normal orbit. Additionally the patients were examined for clinical signs of facial asymmetry and photodocumented.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Segmentation of the right orbit. Grey areas compare with identical thresholds (min − 204 H/max +166 H). Stippled area outlined corresponds to the bony orbital volume. White dots (arrow) belong to the orbital implant.

For quantitative analysis relative volume differences between anophthalmic and normal orbit were calculated:

(BOV_enuc − BOV)/BOV

Non-parametric methods were used for analysis because of smallish numbers involved. The sign test was conducted to assess whether the average difference between the anophthalmic and normal orbit was statistically significantly different from zero. The Mann–Whitney rank sum test was used to assess whether reductions in different groups of patients were statistically significantly different. To assess association between orbital volume and each age at enucleation and time since enucleation, the Spearman rank correlation coefficient was calculated.

Results

Three dimensional reconstructions and volumetric data were obtained from all 33 patients. The image of the three dimensional reconstructed orbit corresponds to an impression mould of the orbit (Fig 2). Repeated segmentations and volume calculations of the same orbit revealed a variation of plus or minus 0.9%, indicating the high precision of the method.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Bony orbital volume compartment, corresponding to an impression mould of the orbit. Thirty five year old patient, enucleated with primary orbital implant age 3.5 years. BOV_enuc(right): 25.5 ml, BOV: 27.1 ml, BOV difference between enucleated and fellow orbit −5.5%.

These data provide strong evidence (p <0.01) of lower anophthalmic orbit volume than normal orbit volume in all patients with long standing anophthalmia, both in patients enucleated in childhood and in adulthood. In group I there was a median change of −1.7 ml (interquartile range −3.0, −1.5 ml), corresponding a relative volume reduction of 7.0% (5.4%, 9.94%) of the enucleated orbit. In group IIa there was an absolute median BOV difference between the enucleated orbit and the normal orbit of −0.9 ml (−2.13, −0.41 ml), which is a 3.8% relative difference (−7.90%, −1.75%). In both groups the differences were statistically significantly different from zero, the differences between groups were not significant.

There was also a mild, but insignificant, decrease of the BOV in group IIb of −0.4 ml (−1.3, −0.01 ml), corresponding to −1.9% relative difference (−5.03%, 0%). The greatest reduction seen in the four controls was −1.8 ml (Table 1).

The relative orbital volume differences for each patient and controls are shown in Figure 3. There were no differences in BOV reduction between patients with and without primary orbital implants.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Bony orbital volume following enucleation. Relative volume differences, single values. Group I: childhood, long standing, group IIa: adulthood, long standing, group IIb: adulthood, short follow up. Negative values indicating smaller volume in the anophthalmic orbit. *Primary orbital implant.

Median values of all groups are illustrated in Figure4.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Bony orbital volume following enucleation. Relative volume differences, median.

The relation between age at enucleation and BOV is shown in a scatter diagram (Fig 5) for patients in groups I and IIa. We found some evidence (not statistically significant) of a positive correlation between the age at enucleation and the BOV difference (rho = 0.36, p = 0.09). There was evidence, however, of a correlation between the influence of the time interval since enucleation and the BOV difference (rho = −0.56, p = 0.003) (Fig6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Scatter diagram. Age at enucleation versus BOV difference enucleated orbit − normal orbit. rho = 0.36, p = 0.09, Spearman rank order correlation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Scatter diagram. Time interval enucleation to CT examination versus BOV difference enucleated orbit − normal orbit. rho = −0.56, p = 0.003, Spearman rank order correlation. Grey diamond = patients enucleated in childhood.

Clinically none of the patients showed signs of facial asymmetry. This is even true for the patient with the largest relative volume difference of −14.7% (Fig 7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

Fifty five year old patient enucleated at age 3 years, after secondary orbital implant. BOV_enuc: 20.2 ml, BOV: 23.7 ml, BOV difference: −14.7%. No clinically obvious orbital malformation or facial asymmetry.

Discussion

“Among the apparent truisms in ophthalmology, there is one, at least, which requires close scrutiny. This is the belief that orbital growth is checked following enucleation in childhood, resulting in more or less maldevelopment and facial asymmetry depending upon the age at which enucleation was performed.” With these sentences Howard begins his article on orbital growth after unilateral enucleation in childhood.10 However, in a series of 15 patients enucleated under the age of 12 years he did not find a significant decrease of the orbital size or an asymmetry. He concluded that enucleation of one eye in infancy or childhood is associated with a small reduction in the size of the bony orbit in later years, but that this reduction is not of cosmetic consequence. The same opinion had been expressed by Gordon Byers from Moorfields Eye Hospital in 1898, based on his study on 10 patients enucleated during childhood.11 Following his report he detected no arrest of development. His observations were questioned by others.1213 Thomson12 carried out experimental enucleations in young rabbits. His results have been considered to show conclusively that, in rabbits, a deficiency of growth occurs in the anophthalmic orbit. In the 1960s Kennedy5 and Sarnat6 began their extensive studies on the effect of early enucleation on the orbit. They found significant reductions of the orbital size particularly in the animal model. Based on approximate estimations, Kennedy5calculated the decrease in the orbital volume up to 19.5% in patients with orbital implants and up to 27.7% in patients without orbital implants. He assumed that the volume reduction after enucleation can reach the magnitude of 35–50%. Their work was of fundamental importance to the understanding of the problem of early enucleation and persistently influenced the point of view of ophthalmologists. Up to now it has been generally accepted that early enucleation causes harmful maldevelopment of the enucleated orbit.14-20

Can our results on three dimensional reconstructions and orbital volumetry in anophthalmic patients add any new aspects to this thoroughly discussed issue? Yes, because for the first time exact volumetric data on a larger series of anophthalmic humans in vivo with very long follow ups are available. Our volumetric studies based on high resolution CT scans confirm a reduction of the anophthalmic bony orbit in patients enucleated in their early childhood. However, the volume reduction is smaller than generally assumed. In our patients we found that the greatest reduction of the anophthalmic orbit compared with the fellow orbit was −14.5% which is less than the few volumetric data given in the literature, where volume reductions were calculated at between 19.5% and 50%.521

Interestingly, we also found a statistically significant reduction of the BOV in patients who were enucleated when they were grown up. In this adult group the volume deficit reached up to 4 ml and the median reduction almost −4% in the enucleated orbit 12 years after enucleation. The adult bony orbit has been regarded as a stable compartment not being influenced by enucleation.1822-25This is obviously not the case. However, there are some early observations suggesting that enucleation might induce shrinkage of the bony orbit2627 which had not yet been proved.

Correlation analysis of BOV changes versus age at enucleation and time interval since enucleation shows a stronger association between reduction of BOV and the time course than age at surgery.

There was no difference between patients with and without primary orbital implants, suggesting that the post-enucleation bony orbital development in humans may not be influenced by implants. Small numbers (4/4) make statistical analysis not applicable. However, in none of our patients was clinically obvious facial asymmetry noticed. This is consistent with reports in the literature on patients who had been enucleated during childhood and who showed no significant facial malformation despite differences in radiological features.1022 Recently Fountain et al examined nine patients 5–10 years after enucleation during early childhood (age 0.4–6 years). Morphometric measurements of five linear orbital dimensions using a millimetre rule were collected. They found no statistically significant difference in any of the measured orbital dimensions between the enucleated and fellow orbits, which was attributed to the presence of an orbital implant.28However, their observation, which is consistent with our clinical impression, is more understandable because of the small volume reductions, which cannot be detected by simple and rough linear measurements than the fact, that all nine patients had primary orbital implants. In our series at least there is no evidence that orbital implants influence BOV after early enucleation.