British Journal of 1995; 79: 457-461 457 Ocular findings in Saudi Arabian patients with

sickle cell disease Br J Ophthalmol: first published as 10.1136/bjo.79.5.457 on 1 May 1995. Downloaded from

Selwa Al-Hazzaa, Alan C Bird, Andreas Kulozik, Beryl E Serjeant, Graham R Serjeant, Peter Thomas, Andrew Padmos

Abstract obvious environmental differences, are pre- Aim-This study was set up to determine sumed to reflect these genetic differences. whether or not retinal changes occur in There are no reports of retinal involvement sickle cell disease in Saudi Arabian in sickle cell disease in Saudi Arabia. subjects with either the Benin, which Observations from general clinical features exists in the south western part of the suggest that patients from the south west might kingdom, or Asian haplotypes in the east, have sickle cell similar to that and to compare the findings with those in observed in patients of African origin, whereas sickle cell disease in Jamaica. those in the east with both high HbF levels and Methods-Retinal examination and frequent a thalassaemia should be protected were performed from vaso-occlusion and consequently show in 61 patients with SS disease (40 eastern, little retinal damage. The present study had 20 south western, 1 central region) and 10 two objectives: firstly, to learn whether sickle with sickle cell ,BO thalassaemia. cell occurs in Saudi Arabia; and, Results-Peripheral retinal vascular secondly, to use the opportunity of the genetic changes were common, and a qualita- and haematological diversity of Saudi patients tively abnormal vascular border believed with sickle cell disease, to learn more about the to imply risk of proliferative sickle risk factors for retinal involvement. retinopathy (PSR) was significantly more common in south western SS patients and PSR was shown in one of these. There Patients and methods were no differences in visual acuity, the presence of peripheral retinal patches, or PATIENTS the circumferential or posterior extent of The patients attended the sickle cell clinic peripheral retinal vessel closure between operated by the Department of Oncology at

King Faisal Specialist http://bjo.bmj.com/ Hospital and Research SS disease and sickle cell PO thalassaemia King Faisal Specialist Hospital, Riyadh, Saudi Center, Riyadh, Saudi or between SS disease in the two regions. Arabia. This clinic serves patients resident in Arabia Compared with the Jamaican Cohort the Riyadh area but also conducts biannual Division of Study, >1800 of the peripheral retinal assessments of patients from both the eastern Ophthalmology, Department of vasculature was seen significantly less and south western provinces where sickle cell Surgery frequent, suggesting factors inhibiting disease is relatively common. The study was S Al-Hazzaa vascular remodelling in Saudi patients in confined to patients showing only haemo- on on September 30, 2021 by guest. Protected copyright. Department of early life. globins F, S, and A2 haemoglobin electro- Oncology Conclusion-Sickle cell disease in Saudi phoresis - that is, homozygous sickle cell (SS) A Padmos Arabia affects the and represents a disease or sickle cell PO thalassaemia (S,O threat to vision. occur thalassaemia), aged 17 years and above who Institute of potential Changes Ophthalmology, whatever the haplotype, and is similar to had fluorescein angiography and angioscopy Moorfields Eye that observed in Jamaica. allowing description of the peripheral retinal Hospital, London (BrJ Ophthalmol 1995; 79: 457-461) vascular border. A total of 88 patients had A C Bird been previously characterised and all were Universittitskliniklum given appointments during a 5 day period in Charit6, Medizinische Sickle cell disease in Saudi Arabia presents April 1993. Seventy four (84%) attended but Fakultilt, Department in Paediatrics, interesting and instructive differences from three SS patients were excluded on the basis of Schumannstrasse that observed in populations of African origin. penetrating ocular trauma (one), high 20/21, 10098, Berlin The mutation in the south west ofthe kingdom (one), and no veins for fluorescein angiography A Kulozik is ofthe Benin haplotype, is probably imported (one). The study group comprised 71 patients, MRC Laboratories from Africa, and has similar levels of fetal 61 with SS disease (33 male) and 10 with SO (Jamaica), University haemoglobin (HbF) to those observed in thalassaemia (five male). The patients were ofthe West Indies, Kingston, Jamaica African patients with the same haplotype.1 In further divided according to geographical B E Serjeant the eastern province, there has probably been origin, 44 (40 SS, four S,B thalassaemia) G R Sergeant an independent mutation of the sickle cell originating in the eastern province, 25 (20 SS, P Thomas gene, the so-called Asian haplotype,2 which five SO thalassaemia) from the south west, Correspondence to: is associated with higher levels of HbF.3 In and two (one SS, one S,O thalassaemia) from Professor Alan C Bird, Professorial Unit, Institute of both areas, the disease coincides with high the central region. Data from these last two Ophthalmology, Moorfields frequencies of a thalassaemia. There are patients were used in the genotype comparison Eye Hospital, City Road, differences in haematological and clinical but the SS subject from the central region was London EC1V 2PD. features of SS disease between the two excluded from the Accepted for publication regions analyses comparing features 10 January 1995 of Saudi Arabia' which in the absence of of the eastern and south western regions. 458 Al-Hazzaa, Bird, Kulozik, Serjeant, Thomas, Padmos

The retinal vascular border as recorded by fluorescein angiography was graded according to a recent classification5 which appears to have prognostic significance. A type I border Br J Ophthalmol: first published as 10.1136/bjo.79.5.457 on 1 May 1995. Downloaded from was qualitatively similar to normal with a smooth border formed of arteriovenous loops, the capillaries becoming longer and less densely packed as the border is approached. The border may differ from normal by being more posteriorly placed, and vascular anomalies intemal to the border are more common than normal. A type II border is irregular, demonstrates a sharp demarcation between perfused and non-perfused retina with abrupt terminations of small or medium calibre vessels, and a dense capillary bed up to the margin of perfusion. This pattem was further subdivided into those manifesting capillary buds or stumps extending into non- perfused retina (type IIa) and those without (type IIb). In some angiograms, the border was unclassifiable either because it was too peripheral to photograph or the photographs Figure I Fluorescein angiogram of a 1 7-year-old man with SS diseasefrom the eastern region showing occlusion were of poor quality. Examinations were of an artery. There is a hypofluorescent column beyond the performed by the two ophthalmologists (SH, vessels implying a recent event. AB) without knowledge of the geographical origin of the patients.

The age of patients with S,B° thalassaemia (mean 24 5 (SD 4 9) years) did not differ from STATISTICAL METHODS that in SS disease (mean 27-7 (SD 6 4); t test, The x2 test for association was used for p=0 13) and that of eastern SS patients (mean nominal variables when the numbers were 28-4, range 17-44 years) did not differ from large enough. When the variables were those in the south west (26-7, range 17-40 ordered, the association was further investi- years; t test, p=0.32). The sex of SS patients gated using the x2 test for trend. When did not differ between regions, males account- expected frequencies were too small, the ing for 23/40 eastern and 9/20 western patients variables were regrouped to a 2 by 2 table, and respectively (X2 test, p=0.52). the association tested using Fisher's exact test. http://bjo.bmj.com/ Comparison was also made with eye changes Logistic regression was used when it was in the older subjects in the Jamaican Cohort necessary to take other factors into account. Study who have angiographic assessments at Comparison of two means was made using the annual intervals. Of the 43 SS subjects aged t test, with separate variance estimates when over 17 years at their 1992 assessment, the variances were not homogeneous. When there mean age was 17-8 (range 17-0-18-6) years. were more than two means, analysis of variance was used, with Bonferroni adjusted on September 30, 2021 by guest. Protected copyright. p values to test pairwise contrasts. The HAEMATOLOGICAL METHODS distribution of HbF was highly skewed and was The diagnosis of SS disease and of sickle transformed by loge (HbF+ 1) before analysis. cell thalassaemia was based on standard criteria.4 Routine haematological indices were measured in a Coulter STKR (Coulter Results Electronics) and HbA2, HbF, the number of Retinal vascular changes were identified in this at globin genes, and the globin haplotype study which were similar qualitatively to those determined as previously reported.' seen in the Jamaican cohort comprising fresh vascular obstruction (Fig 1), posterior dis- placement of the vascular border, and abnor- OPHTHALMIC METHODS malities of the peripheral vasculature (Figs 2 After assessment of best corrected visual and 3). acuity and pupillary dilatation, the fundus was examined by indirect record- ing the number and position of haemorrhages, COMPARISON OF SS DISEASE AND SICKLE CELL chorioretinal scars, retinal schisis cavities B0 THALASSAEMIA (collectively termed patches), and PSR lesions. There were only 10 patients with sickle r10cell Fluoroscopy and fluorescein angiography were thalassaemia so data from eastern, south performed after the injection of a bolus of 5 ml western, and central regions were pooled for of 20% sodium fluorescein. The extent of comparison with SS disease. The number of peripheral vascular border seen by indirect ao globin genes (known in 53 SS, six S0 ophthalmoscopy was recorded and the border thalassaemia) did not differ significantly was defined as posterior if it was behind the between genotypes, there being 25 oot/oxao, equator. 20 a-/aao, and eight ao-/a- in SS disease Ocularfindings in Saudi Arabian patients with sickle cell disease 459 Br J Ophthalmol: first published as 10.1136/bjo.79.5.457 on 1 May 1995. Downloaded from

Figure 2 Fluorescein angiogram ofa 29-year-old woman with SS diseasefrom the eastern region showing a type IIb Figure 3 Fluorescein angiogram ofa 23-year-old man border. with SS diseasefrom the western region showing a type IIa border.

compared with two, three, one for SB° thalas- REGIONAL DIFFERENCES IN OPHTHALMOLOGY saemia. There were no differences in presence OF SS DISEASE of patches, degree of circumferential closure, Visual acuity was <6/12 in seven (18%) or posterior closure between different geno- eastern patients and in three (15%) of south types. Patches occurred in 26 SS and three 813 western patients (p= 1 00 using Fisher's exact thalassaemia (Fisher's exact test, p=0.70), test). Acuity was <6/60 in three (4%) eyes of > 1800 ofthe vascular border was seen in 33 SS eastern subjects and in two (5%) eyes of south and two S1O thalassaemia (Fisher's exact test, western subjects none of which was attribut- p=0093), and posterior position occurred in able to sickle cell disease. Patches occurred in 18 SS and two S1O thalassaemia (Fisher's exact the peripheral retina of 16 (40%) eastern test, p=084). The angiographic description of patients and in 10 (50%) of south western retinal vascular border using a summary score patients, this difference not reaching signifi- ofthe worst affected eye did not differ between cance (X2 test, p=0 65). genotypes, type II borders occurring in 15/58 In all patients the peripheral border of http://bjo.bmj.com/ SS and in 0/10 813 thalassaemia (Fisher's retinal vascularisation was seen, the circum- exact test, p=0 13). In summary, although the ferential extent being > 1800 in 20 (50%) small sample of subjects with sickle cell 130 eastern patients and in 12 (60%) south western thalassaemia renders the tests relatively in- subjects (X2 test, p=0 65). Total circumferen- sensitive, there did not appear to be major tial closure occurred bilaterally in 19 (48%), differences from SS disease. unilaterally in one eastern patient, and bilater- ally in 12 (60%) south western patients. on September 30, 2021 by guest. Protected copyright. Posterior closure occurred in 11 (28%) eastern REGIONAL DIFFERENCES IN HAEMATOLOGY OF and in six (30%) south western patients, there SS DISEASE being no regional difference (X2 test, p= 1 00). The number of ao globin genes was known in Classification of the retinal border on 33 eastern and 19 south western subjects, there fluorescein angiography was possible in 71 being eight homozygotes for a thalassaemia (89%) eyes of eastern and in 38 (95%) eyes of (ao-/a-), 13 heterozygotes (a-/ao), and 12 south western patients (Table 1). Border type with a normal a globin gene complement IIa was significantly more frequent in south (aa/aao) in the east compared with 0, six, and western patients (Fisher's exact test, p=003). 14 in the south west. The trend of an increas- Proliferative sickle retinopathy occurred in one ing east:west ratio with decreasing number ofa subject, a 35-year-old man from the south globin genes was statistically significant west. (p<0-0 1). Total haemoglobin levels in eastern patients (mean 10-5, SD 1-5) were significantly higher than in the south west (9.4, 1 5) (t test, EYE VARIABLES AND HAEMATOLOGY p=0-0 13). Fetal haemoglobin levels in eastern The regional differences in haematology patients (median 12-0%, range 3-3-22 1) were imply that analysis of the relation between eye also significantly higher than in the south west variables and haematology should take region (median 4 440/o, range 0-5-17T9) (t test on into account using analysis of variance. transformed data, loge (HbF+ 1), with separate Subjects with >1800 of the anterior border variance estimates, p<0 001). Mean cell seen had a lower HbF compared with volume in the eastern patients (mean 83-0 fl, those with - 1800 (Table 2). The relation SD 11.1) was significantly lower than in between the extent of the border seen and a south western patients (mean 89-7 fl, SD 10-4) thalassaemia (heterozygous and homozygous (t test, p=0 03). forms combined) appeared to differ in the two 460 Al-Hazzaa, Bird, Kulozik, Serjeant, Thomas, Padmos

Table 1 Angiographic classification of the peripheral retinal vascular border in SS Discussion patients from the two regions ofSaudi Arabia. Each eye is scored as to the most serious characteristic, and the summary score refers to a subject scored as to the eye with the most Despite the small number of patients in this serious characteristic study it is clear that SS disease in Saudi Arabia is associated with peripheral retinal vessel Br J Ophthalmol: first published as 10.1136/bjo.79.5.457 on 1 May 1995. Downloaded from Eastern region (%o) Western region (%o) changes which are similar to those seen in the Border Left Right Summary Left Right Summary subjects of African origin. The finding that 10 type eye eye score eye eye score (14%) of patients had sickle cell PO thalas- I 33 (87) 28 (85) 31 (82) 13 (68) 12 (63) 12 (63) saemia among 71 subjects with an SS pheno- IIa 1 (3) 0 1 (3) 5 (26) 4 (21) 5 (26) type, suggests a greater relative frequency of Ilb 4 (11) 5 (15) 6 (16) 1 (5) 3 (16) 2 (11) Total 38 33 38 19 19 19 this genotype than occurs in populations of African origin. The retinal involvement in sickle cell PO thalassaemia was similar to that in regions. There was no association in the east- SS disease although no patients with S0 ern province, ao thalassaemia occurring in thalassaemia had a type II border in this study. 10/15 subjects with <1800 compared with It was of interest to assess whether retinal 11/18 subjects with >1800 (p=1 00 using vascular involvement differed between the two Fisher's exact test). In south western patients, regions since the Benin haplotype occurs in the at thalassaemia occurred in 0/7 with 6 1800 and south west and the Asian haplotype accounts in 6/12 with > 1800 (Fisher's exact test, almost exclusively for the disease in the eastern p=004), suggesting at thalassaemia as a risk province, and eastern patients showed a factor for posterior displacement of the vas- greater frequency of at thalassaemia, higher cular border. total haemoglobin and HbF levels, and lower Patches were associated with lower HbF and values of mean cell volume. The type Ila lower mean cell volume. There were no associ- border, which indicates a threat of proliferative ations between age, haematology, and number retinopathy, was more common in south of at globin genes with either posterior closure western patients among whom was the only or border scores (Table 2). case of proliferative disease. However, the presence of patches in the peripheral retina, or peripheral retinal vessel change as determined COMPARISON WITH JAMAICAN DATA by either circumferential or posterior extent When compared with subjects in the two was not different. regions of Saudi Arabia, Jamaican patients The haematological correlates of these retinal had significantly lower HbF (mean 7 9, SD changes were less clear. A low HbF level was 1 2) than either region, lower HbF (median significantly associated with more retinal 4*5, range 0A4-14-9) than eastern subjects patches and with a greater extent of circum- only, similar mean cell volume values (mean ferential retinal closure, those observations 84-8, SD 8 9) to those of both regions and being consistent with the documented in- significantly less at thalassaemia (30 aoat/aaoto, hibitory effect of high HbF levels on polymeri- http://bjo.bmj.com/ 13 a-/faat, 0 at-/a-) than the eastern sation of HbS molecules, sickling, and region. presumed vaso-occlusion. However, the higher The presence of patches (26 (43%) Saudi HbF in eastern patients was not reflected in a patients, 24 (59%) Jamaicans) and of posterior lower prevalence of patches and less retinal closure (17 (28%) Saudi patients, 14 (33%/o) closure as might have been expected. a Jamaican patients) did not differ between the Thalassaemia also inhibits sickling probably groups (X2 test, p=019 and p=0 75 respec- through its effect in lowering the haemoglobin on September 30, 2021 by guest. Protected copyright. tively). The proportion of patients with > 1800 concentration within cells, and in Jamaican of the circumference of the vascular border studies at thalassaemia reduced the risk of seen (32/60 Saudi patients, 39/42 Jamaican retinal closure.6 7 The greater frequency of an patients) was significantly greater in Jamaicans unstable retinal border in south western patients (p<0 001 using x2 test). Comparison of angio- was not clearly attributable to the lower graphic border type showed that type IIa was frequency of at thalassaemia or the lower HbF significantly more common in south western levels in these patients, but because of the small patients (5/19) than in Jamaicans (1/36) numbers, these analyses were relatively insensi- (Fisher's exact test, p=0 03), there being no tive. It was of interest that the high frequency difference between eastern Saudi patients and of at thalassaemia in the eastern province did Jamaicans (Fisher's exact test, p= 1 00). not appear to protect against vascular changes.

Table 2 Relation between ocularfindings (3180 ofthe vascular border seen, presence ofpatches, poste?ior position ofthe vascular border, and border classified as II) and age and haematological indices

Adjusted mean differences (95% confidence interval) Adjusted odds ratio (95% CI) for Age Hb loge (HbF+ )* MCV a thalassaemia Vascular border seen -0-9 (-4-3, -2 4) -0 1 (-0 9, 0 7) W -03 (-1-0, 0 4) -3-2 (-8-8, 2 5) See text E -0 3 (-0 5, 0 0) Patches -2-2 (-5-5, 1-2) -0-2 (-1-0, 0 6) W-0-6 (-1,2, 0 1) -7-5 (-12-9 -2-1) 2-6 (0 7, 9 1) E -0 3 (-0 5, 0°0) Posterior border -0 4 (-4-1, 3 4) 0 7 (-0-2, 1-5) W 0 3 (-0 4, 1 0) -2-1 (-8-4, 4-3) 1 0 (0 3, 3 8) E -0-1 (-0 4, 0 2) Border classification 0 0 (-0 4, 3-9) 0-7 (-0-3, 1-6) W 0 0 (-0 7, 0 7) 2-5 (-4 4, 9 4) 1-2 (0 3, 5-1) E -0-1 (-0 5, 0 2)

*Variance differed between regions so regions presented separately. MCV=mean cell volume. Ocularfindings in Saudi Arabian patients with sickle cell disease 461

Comparison with the Jamaican cohort study which may otherwise be relatively mild. should be treated with caution because of the Although no case in this study had lost vision difference in mean ages (27-7 years for Saudi as a consequence of their haemoglobinopathy, patients, 17-8 years for Jamaicans), and the the number of cases was small, and it is likely Br J Ophthalmol: first published as 10.1136/bjo.79.5.457 on 1 May 1995. Downloaded from Saudi patients were not derived from un- that SS disease represents a threat to vision in selected population base. In Jamaican subjects Saudi Arabia albeit small. The more florid reti- the extent of vascular border seen was greater, nal pathology often observed in Jamaican despite their younger mean age, suggesting a patients is generally attributable to sickle cell factor inhibiting vessel closure in both regions haemoglobin C (SC) disease or sickle cell P+ of Saudi Arabia in early life.5 This factor is thalassaemia which were not encountered in unlikely to be the frequency ofhigh HbF levels Saudi Arabia.5 or ox thalassaemia since these did not differ We thank Dr Gudrun Brismar, head of the division of oph- between south western Saudi and Jamaican thalmology, Department of Surgery for permission to use the patients. Furthermore, the type IIa border was facilities of that department, and Dr Muir Jackson and his staff in the haematology section, Department of Pathology and more common in south western than Jamaican Laboratory Medicine for laboratory facilities. Mr Khalid Al patients, possibly reflecting the greater mean Abdulkareem, Mr Mohammed Dokmak, Mr Ahmed Aba Wazeid, and Mrs Fawzia El Elimy provided administrative age of south western Saudi patients, a specific assistance. risk factor, or the symptomatic bias inherent in patients attending a tertiary referral hospital. 1 Padmos MA, Roberts GT, Sackey K, Kulozik A, Bail S, Morris JS, et al. Two different forms of homozygous sickle The greater prevalence of an unstable type IIa cell disease occur in Saudi Arabia. BrJ7 Haematol 1991; 79: border may be relevant for visual prognosis 93-8. 2 Kulozik AE, Wainscoat JS, Serjeant GR, Kar BC, Al-Awamy since this border implies threat of PSR which B, Essan GJF, et al. Geographical survey of ,B-globin gene may in turn lead to visual loss. By contrast, a haplotypes: evidence for an independent Asian origin ofthe sickle-cell mutation. Am 7 Hum Genet 1986; 39: 239-44. type I border, however posteriorly placed, is 3 Pembrey ME, Wood WG, Weatherall DJ, Perrine RP. Fetal not followed by this and may be haemoglobin production and the sickle gene in the Oases of Eastern Saudi Arabia. BrJ Haematol 1978; 40: 415-29. protective of proliferative disease.5 4 Serjeant GR. Sickle cell disease. 2nd ed. Oxford: Oxford Overall, despite minor differences, the University Press, 1992. 5 Penman AD, Talbot JF, Chuang EL, Thomas P, Serjeant ocular pathology in SS patients from each GR, Bird AC. A new classification of peripheral retinal region of Saudi Arabia and differed little from vascular changes in sickle cell disease: relevance to sight threatening disease. Br J7 Ophthalmol 1994; 78: that observed in Jamaicans with SS disease. 681-90. That this is the case despite differences in 6 Talbot JF, Bird AC, Rabb LM, Maude GH, Serjeant GR. Sickle cell retinopathy in Jamaican children - a search for haematological variables and clinical features prognostic factors. BrJ Ophthalmol 1983; 67: 782-5. should not be surprising, since visually 7 Talbot JF, Bird AC, Maude GH, Acheson RW, Moriarty BJ, Serjeant GR. Sickle cell retinopathy in Jamaican children: threatening retinal vascular disease is seen further observations from a cohort study. Br J Ophthalmol most commonly in forms of sickle cell disease 1988; 72: 727-32. http://bjo.bmj.com/ on September 30, 2021 by guest. Protected copyright.