Article Text
Abstract
Aim To evaluate the histopathologic factors statistically associated with the presence of calcification in eyes with retinoblastoma.
Methods Retrospective, consecutive and observational case series. Three hundred and two enucleated eyes with retinoblastoma examined between the years 1960 and 2008. Five representative histopathologic slides of the pupil–optic nerve section and three cross optic nerve sections were retrospectively reviewed. The presence and degree of calcification as well as other histopathologic features were evaluated. Demographic data including age, gender and country of origin of the case were also reviewed. Univariate and multivariate statistical analyses were performed to search for a possible correlation between calcification and the other histopathologic factors and/or demographic data.
Results Calcification was present in 84.9% of cases. Age, tumour size, necrosis, basophilic staining, iris neovascularisation, choroidal, scleral and/or optic nerve invasion were correlated significantly with calcification. Multivariate analysis showed a significant correlation between the presence of calcification and the amount of necrosis and choroidal invasion only.
Conclusions In this series, calcification was more frequent in cases with more necrosis and cases with choroidal invasion, a known poor histopathologic risk factor for metastatic disease. The possible clinical implication of the findings from this study deserves additional studies.
- Retinoblastoma
- calcification
- histopathology
- statistical analysis
- pathology
Statistics from Altmetric.com
Introduction
Retinoblastoma is the most common intraocular tumour in children. Metastatic disease is still associated with a poor prognosis.1 Histopathology of enucleated cases is essential to evaluate possible risk factors for metastatic disease, such as the presence of tumour in the optic nerve posterior to the lamina cribrosa, tumour at the surgical edge of the optic nerve, massive choroidal invasion and scleral and orbital invasion.1–3
Calcification in retinoblastoma is a frequent histopathologic finding with a reported frequency up to 95% of cases, and possibly more common in those lesions undergoing spontaneous or induced regression,4–6 although the subject has not been studied in depth. There are two forms of pathologic calcification.7 Dystrophic calcification is mostly encountered in areas of necrosis, as in retinoblastoma; it occurs despite normal serum levels of calcium and in the absence of derangements in calcium metabolism. By contrast, metastatic calcification occurs in otherwise normal tissues and it almost always results from hypercalcaemia secondary to some disturbance in calcium metabolism.
In previous reports on histopathologic features and/or prognostic factors in retinoblastoma cases, either there were no data on calcification at all or there was no correlation with other factors. To the best of our knowledge, no previous study has focused on the calcification in retinoblastoma and tried to address the relationship between calcification and other prognostic, histopathologic or epidemiologic factors, including the poor prognostic factors for metastatic disease. The purpose of the present study is to evaluate the histopathologic factors statistically associated with the presence of calcification in eyes with retinoblastoma.
Methods
We retrospectively reviewed all the enucleated eyes with retinoblastoma examined at the Ophthalmic Pathology Laboratory at the Hadassah-Hebrew University Medical Center, Jerusalem, Israel, between 1960 and 2008. This study was approved by the Hadassah-Hebrew University Medical Center Institutional Review Board. The specimens were fixed in 10% neutral-buffered formalin for at least 48 h, examined grossly and processed. The pupil–optic nerve (P–O) section and distal (surgical edge) cross optic nerve section were embedded in paraffin; the superior and inferior calottes were processed in selected cases also. The presence and degree of calcification as well as other histopathologic features (size, focality, pattern, penetration into the orbit, nerve, sclera and/or choroid, retina, subretinal, vitreous and/or anterior chamber seeding, iris neovascularisation, Flexner–Wintersteiner rosettes, Homer Wright rosettes, fleurettes, basophilic staining (basophilia of Bruch's membrane, vascular walls and DNA lakes), subretinal seeding, optic atrophy, cholesterol clefts, cataract and angle closure) were evaluated. Figure 1 shows calcification, necrosis and choroidal invasion. More than 15 P–O sections and 5 optic nerve sections were stained with H&E. The following demographic data were collected as available: age at enucleation, gender and country or origin of the patient.
For each case at least full five representative non-consecutive histopathologic slides of the P–O section and three cross optic nerve sections were retrospectively reviewed. All the slides were reviewed by two ophthalmic pathologists (JP, JL). The presence of calcification within the tumour was classified in a semiquantitative fashion into four groups: 0, no calcification; 1, few isolated foci; 2, up to 25% of tumour volume and 3, 26% and more of the tumour volume is calcified.
The following parameters were collected for each case as previously reported by Pe'er et al8: tumour size (graded into four groups according to the eye area occupied by the tumour in the histopathologic slide: 1, ≤25%; 2, 26%–50%; 3, 51%–75% and 4, ≥76% of the eye area); focality (multifocal vs monofocal); pattern (endophytic, exophytic, mixed, diffuse, total necrosis); orbital invasion (yes/no); choroidal invasion (divided into five groups: 0, no invasion; 1, superficial penetration into the choriocapillaries; 2, full focal penetration; 3, up to 50% of the choroid is occupied by tumour and 4, more than half of the choroid is occupied by the tumour); scleral invasion (divided into five groups: 0, no invasion; 1, superficial scleral invasion; 2, up to half of scleral thickness; 3, more than half of the scleral thickness is invaded by tumour and 4, the tumour invades through the sclera); optic nerve invasion (divided into five groups: 0, no invasion; 1, prelaminar; 2, laminar; 3, retrolaminar and 4, invasion at the surgical edge); retina, subretinal, vitreous and anterior chamber seeding were recorded (yes/no); presence of Flexner–Wintersteiner and Homer Wright rosettes and fleurettes (yes/no); number of mitoses (in 20 high power fields); amount of necrosis (percentage of total tumour area using an ocular micrometre grid with 10×10 divisions); basophilic staining (yes/no); iris neovascularisation (divided into five groups: 0, no iris neovascularisation; 1, fine isolated capillaries with the normal iris structure remaining intact; 2, a layer of thin capillaries; 3, a prominent vascular membrane with flattening of the iris surface with or without ectropion uvea and 4, a dense fibrovascular membrane covering the whole iris); presence of optic atrophy, cholesterol clefts, cataract and histopathologic angle closure (yes/no) and millimetres of optic nerve resected.
Statistical methods
The outcome variable (target end point) was the dichotomous indicator of any calcification.
Univariate analysis
In univariate preliminary analyses, the association of each of the demographic and histological categories was tested by means of a χ2 or Fisher's exact test (see tables 1 and 2).
The multivariate model
First a multivariate model was built including all variables statistically significant in the univariate step at the <0.10 level. It included age, gender, laterality, size, pattern, penetration into the orbit, penetration into the nerve, penetration into the sclera, penetration into the choroid, iris neovascularisation, basophilic staining and necrosis. Only 84 subjects had information on all variables (for 218 subjects the information of at least one variable was missing). The variables with the greatest number of missing information were excluded sequentially: laterality (168 cases), pattern (104 cases), gender (77 cases) and age (72 cases). The number of cases remaining for the multivariate analysis rose to 234. Extremely high ORs called for the analysis of multicollinearity. The variables ‘penetration into the sclera’ and ‘size’ with the lowest tolerance were removed from the model. To deal with the interaction between ‘penetration into the orbit’ and ‘penetration into the nerve,’ a composite variable was constructed indicating ‘penetration into the orbit or into the nerve’. A backwards likelihood ratio multivariate logistic regression entering the variables ‘basophilic staining,’ ‘penetration into the choroid,’ ‘iris neovascularisation,’ ‘necrosis’ and ‘penetration into the orbit or into the nerve’ was performed. The final model including only the variables statistically significant was built. The purpose of this last step was to lower the number of exclusions due to missing values.
Statistical analyses were performed using SPSS 15.0 (SPSS Inc.).
Results
Three hundred two cases with retinoblastoma were retrospectively reviewed. Of these 302 eyes, 140 cases (46%) were sent to the laboratory from various African countries, 6 cases (2%) were received from Latin American countries, whereas the remaining 156 eyes (52%) were enucleated in Israel. The average age at enucleation was 23.4 months (SD 34.1 months, median 17 months, range, 0–156 months). Fifty-seven per cent of children were male and the right eye was affected in 49% of cases.
Calcification was present in 259 cases (85.8%). Eighty-four cases (27.8%) were in group 1, 159 cases (52.6%) were in group 2 and 16 cases (5.3%) were in group 3. Of the cases from Africa, 92.9% had calcification, whereas 81.5% Israeli cases had calcification; this difference was statistically significant (p=0.0039).
The histopathologic features are summarised in table 1. The average mitotic activity was 45±37 mitoses per 20 high power fields (range, 0–187). The mean amount of tumour necrosis was 50.7±28.9% of the total tumour area; in 32% of cases more than two-thirds of the tumour was necrotic. Retinal seeding, optic atrophy, cholesterol clefts and cataract were present in 65.9%, 82.9%, 5.7% and 18.7% of cases, respectively. The resected optic nerve measured 6.7±3.5 mm (range, 0.5–14.0 mm, median 7.0 mm).
The possible association between calcification and the rest of histopathological features and epidemiological data was then analysed. Univariate analyses considering three calcification subgroups (group 0, group 1 and group 2 or 3) demonstrated a statistically significant association between calcification and age (p=0.001), gender (p=0.02), tumour size (p<0.0001), focality (p=0.01), tumour pattern (p=0.0004), orbital invasion (p<0.0001), optic nerve invasion (p=0.0002), scleral invasion (p=0.001), choroidal invasion (p<0.0001), iris neovascularisation (p=0.01), basophilic staining (p<0.0001), amount of necrosis (p<0.0001) and cholesterol clefts (p=0.01). Results are shown in table 2. Association with choroidal invasion was also found to be significant when cases were grouped by their origin (Israel vs Africa, likelihood ratio p=0.034 and p=0.031, respectively).
Univariate analyses considering only two calcification subgroups (no/yes) demonstrated a statistically significant association between calcification and age (p=0.001), tumour size (p<0.0001), tumour pattern (p=0.003), orbital invasion (p=0.03), choroidal invasion (p<0.0001), basophilic staining (p<0.0001) and amount of necrosis (p<0.0001). However, gender (p=0.07), focality (p=0.32), optic nerve invasion (p=0.09), scleral invasion (p=0.07), iris neovascularisation (p=0.11) and cholesterol clefts (p=0.34) were found not statistically significantly associated with calcification (table 3), as opposed to the first univariate analysis performed.
Finally, multivariate analysis demonstrated only two histopathologic factors significantly associated with calcification: choroidal invasion (OR=3.26, p=0.002) and amount of necrosis (OR=1.02, p=0.004). Data are shown in table 4.
Discussion
Radiologic foci of calcification are very characteristic but not exclusive of retinoblastoma and an extremely important feature when evaluating a child with suspected retinoblastoma.9 Reported rates of calcification in imaging studies (CT and MR scans) are as high as 95% of retinoblastoma cases.10 11 However, less importance has been given to histopathologic calcification. In the large series reporting histopathologic features and/or prognostic factors, either there are no data on calcification at all or there is no correlation with other factors. To the best of our knowledge, no previous study has focused on the calcification in retinoblastoma and tried to address the relationship between retinoblastoma and other prognostic, histopathologic or epidemiologic factors. Statistical studies on general histopathologic features of retinoblastoma in smaller case series than in the present study, have included calcification in their results, but have not focused exclusively on the statistical association of calcification and the other histopathologic features.12 13
Calcification in retinoblastoma is a frequent histopathologic finding and possibly more common in those lesions undergoing spontaneous or induced regression.4–6 Calcification is also present in retinocytoma which is no longer believed to be a lesion that has undergone spontaneous regression, but rather a benign tumour arising from a retinal cell or a retinoblastoma precursor.14 15 In retinocytoma, calcification occurs within viable parts of the tumour, not in areas of necrosis like the typical calcification in retinoblastoma. Other types of intraocular calcification that occurs within the eye are the basophilic, multilaminated calcospherites in the retinal astrocytic tumours of tuberous sclerosis complex.16 And finally the most common form of intraocular calcification is osseous metaplasia of the retinal pigment epithelium in phthisical eyes with chronic retinal detachment.17
Lin and Tso18 studied calcification in three retinoblastoma cases (two after irradiation and one after spontaneous tumour necrosis) by electron microscopy. They found intramitochondrial calcium deposition after plasmalemma disruption. With progressive necrotic liquefaction of tumour cells—a central calcified cone surrounded by an electron-translucent zone with coarse needle-like calcium deposition on the surface appeared.18 Calcium deposition is possibly more common in those retinoblastomas undergoing spontaneous or induced regression.6 In the present study in univariate analyses considering calcification in a dichotomous way, we found a statistically significant positive association between calcification and age, tumour size, orbital invasion, choroidal invasion, basophilic staining and amount of necrosis. More calcification was found in older children, large tumours, cases with orbital or choroidal invasion and cases with basophilic staining and more necrotic tumours. Part of these findings is in consonance with the already known facts: calcification is more frequent in necrotic cases; tumours in older children may undergo regression or necrosis more frequently and larger tumours have more necrotic areas and DNA released from necrotic cells is responsible for the basophilic staining. Extensive ocular tissue and tumour necrosis has been recently associated with histopathologic high-risk prognostic factors for tumour metastasis and mortality,19 although in that study the incidence of calcification was not commented. However, the possible association between orbital and choroidal invasion (accepted poor prognostic factors for metastatic disease) and calcification has not been reported to date. We found more calcification in cases with more and deeper choroidal invasion. Although it was beyond the scope of this study, the possible association between histopathologic or even radiologic calcification in retinoblastoma and prognosis merits further evaluation. Finally, multivariate analysis demonstrated only two histopathologic factors significantly associated with calcification: choroidal invasion and amount of necrosis.
Histopathologic evaluation of all the parameters was done after revision of at least five not consecutive P–O sections. Calcification was classified in three categories. The subgroups for choroidal invasion were previously reported by our group,8 and are different from other proposed criteria not yet validated.20
The major limitation of our study is that we could not include clinical data for evaluation. Almost half of the cases (48%) were from other countries and no clinical information was available. This was the main reason for us to first undertake a histopathologic study exclusively. The possible influence of previous treatment on the amount of necrosis and calcification21 could not be ascertained. However, our study group was large enough to obtain statistical significance repeatedly when considering all the subgroups of choroidal invasion and calcification, and also when comparing one or the two groups in a dichotomous way, and the results were confirmed by the model we built for multivariate analysis. The other limitation of the study is that due to technical reasons we could not evaluate all the P–O and calottes slides for every case so hypothetically we could miss cases with choroidal invasion and to some extent extrascleral extension in those cases in which the calottes were not reviewed. It is not a routine proceeding in these moments in the management of a retinoblastoma case. The possible future benefits of this specific handling are being evaluated in a prospective trial20; should they prove to be important, the routine processing of retinoblastoma specimens should change accordingly.
In conclusion in our series, we found that calcification was more frequent in cases with more necrosis and cases with choroidal invasion. Additional studies are needed to elucidate whether calcification in retinoblastoma is just a secondary event without importance related to necrosis or can be looked as a poor prognostic factor per se.
References
Footnotes
Presented as a poster at the Association for Research in Vision and Ophthalmology annual meeting, Fort Lauderdale, Florida, USA, May 2010.
Competing interests None.
Ethics approval This study was conducted with the approval of the Hadassah-Hebrew University Medical Center.
Provenance and peer review Not commissioned; externally peer reviewed.