Spleen size in homozygous sickle cell disease: trends in a birth cohort using ultrasound

Abstract

Objectives:

To provide ultrasound baselines for spleen length in homozygous sickle cell disease (HbSS) and in normal controls with a HbAA genotype.

Methods:

The Jamaican cohort study identified 311 babies with HbSS and 246 matched HbAA controls during the screening of 100,000 consecutive deliveries in Kingston, Jamaica from 1973 to 1981. Ultrasonography commenced in 1988 when the youngest patients were aged 6 years at which time deaths, emigrations and default had reduced the numbers to 206 HbSS and 89 controls. It continued annually until 2000.

Results:

The spleen was visualized in all HbAA controls but in only 1103/2138 (52%) scans in HbSS. Where available, mean splenic lengths were significantly lower in HbSS (77–103 mm in males, 70–83 mm in females) compared to normal controls (89–101 mm in males, 86–95 mm in females). Assessed by statistical modelling after adjusting for body height, the splenic ratio (splenic length/body height) declined over the age range 12–20 years in HbSS, consistent with progressive splenic fibrosis. Genetic factors known to inhibit sickling, α thalassemia and fetal hemoglobin level (HbF) significantly reduced the decline in splenic ratio. Clinical splenomegaly was an insensitive measure of splenic enlargement as only 50% of patients aged 18 years and above with spleens measuring ≥150 mm on ultrasonography had palpable spleens.

Conclusions:

An age-related decline in splenic length occurred in HbSS and occurred more slowly with genetic factors known to inhibit sickling. The standards provided may be of value in assessing minor degrees of subclinical acute splenic sequestration.

Advances in knowledge

These are the first standards available for splenic length in HbSS. They may be useful in detecting red cell sequestration, not apparent from clinical splenomegaly and also provide a model for identifying factors inhibiting vaso-occlusion.

Introduction

The spleen is central to much of the early pathology of sickle cell disease as high levels of sickle hemoglobin (HbS) reduce the pliability of red cells which occlude the splenic circulation. This process is most marked in homozygous sickle cell disease (HbSS) in which the spleen becomes clinically enlarged in the first few months of life ¹ and gradually loses function ² rendering patients prone to invasive bacteria. ³ Superimposed on this pattern, some patients develop acute enlargement trapping red cells (acute splenic sequestration) or sustained chronic enlargement (chronic hypersplenism). In the Jamaican cohort, acute splenic sequestration predominantly affected children under 5 years, was a common cause of early death ⁴ and mortality was reduced by teaching mothers splenic palpation and by splenectomy in recurrent cases. ⁵ Chronic hypersplenism usually affects older children, and is associated with sustained rapid hemolysis but the low hemoglobin and the demands of the expanded bone marrow compete with those for growth ^6,7 and patients often require repeated transfusion or splenectomy. The present study seeks to determine baselines in splenic size during the steady state and does not include patients with acute sequestration although may include mild hypersplenism. The routine clinical practice of palpating the spleen in sickle cell disease is an inaccurate indicator since it may not become palpable until 2–3 times its normal volume. ^8,9 Splenic length, as measured by ultrasonography, correlates well with splenic volume determined by CT, ¹⁰ and may be useful in assessing splenic pathology in sickle cell disease. Despite a wealth of observations on ultrasonography in other medical conditions, there are no standard values for patients with sickle cell disease. The Jamaican Cohort Study, ¹¹ in which a large group of patients have been followed from birth provides a unique opportunity to provide such standards, and the results are presented.

Methods and materials

Patient ascertainment

The Jamaican Cohort Study is based on 311 cases of HbSS detected during the screening of 100,000 consecutive non-operative deliveries at the main Government Maternity Hospital (Victoria Jubilee) in Kingston, Jamaica, between June 1973 and December 1981. ¹¹ The first 125 patients with HbSS were gender-matched with two controls with a normal HbAA genotype one born immediately before and one after each index case, matching usually within minutes of birth time. Of these 250 controls, 4 were subsequently found to have the β thalassemia trait leaving 246 (HbAA) controls. Logistical considerations limited the control group but screening continued to 100,000 births so recruitment of babies with HbSS continued for longer (birth dates July 8, 1973–November 5, 1981) than the AA controls (July 7, 1973–January 16, 1976) with resulting differences in age range. Patients and controls were followed prospectively with monthly assessments to 6 months, alternate months from 6 to 12 months, 3-monthly to 5 years, and 6-monthly thereafter. All were requested to attend regularly when clinically well and were encouraged to attend at any time if sick. The social and geographical context of Jamaica is also important as although born in Kingston, Jamaica, the population subsequently disperses throughout the island and some patients had a journey of over 100 km to attend the Sickle Cell Clinic in Kingston where ultrasonography was performed. All services were provided free but patients with limited resources required collection or reimbursement of bus fares.

Patients were invited for ultrasound studies of the spleen, kidneys and gallbladder during 3 week periods each year in January/February over a 13-year period (1988–2000) when the cohort was aged 6–24 years. This timing was made possible by a consultant radiologist from the Royal Berkshire Hospital, Reading (TMW) who donated 3 weeks annually over this period. Of the 311 HbSS patients admitted to the Cohort study, 105 patients (55 died, 16 emigrated, 34 had had splenectomy) were not available for splenic ultrasound and the remaining 206 patients had 2,138 ultrasound scans. Two HbSS patients with hypersplenism had splenectomy during the study, one emigrating after a single assessment in 1988 and the other having six assessments before splenectomy at age 15.8 years in 1993. The study was not designed to observe serial changes in normal controls and there were logistical limitations on the number that could be accommodated but 89/151 (59%) available controls had 128 scans, mostly in 1992/1993.

Laboratory methods

The diagnosis of hemoglobin genotype was made by hemoglobin electrophoresis on cellulose acetate with confirmation of all electrophoretically abnormal samples on agar gel ¹² and supported by compatible HbA₂ and fetal hemoglobin (HbF) values, family studies and DNA sequencing, if parental studies yielded inconsistent results. Two genetic factors known to inhibit sickling, α thalassemia status and HbF level, were examined as possible determinants of splenic length, and the methods of laboratory diagnosis have been described elsewhere. ¹³

Ultrasonography

This was performed with an ATL Ultramark IV machine with a 5 MHz curvilinear probe although two obese children required a 3.5 MHz curvilinear probe. Splenic length was measured in the right lateral decubitus position and scanned in the intercostal plane or subcostally in its long axis.

Statistical methods

Age and sex-specific spleen length (in mm) were tabulated for each genotype (HbAA, HbSS) along with the number of observations and of subjects contributing to these summary values. The distribution of unadjusted values (spleen length) and adjusted values (splenic length divided by participant height in metres) are presented as a smoothed median, with 25th and 75th percentiles, by genotype and gender in 2-year age bands between 6 and 24 years.

Modelling was used to estimate adjusted spleen length between the ages of 12 and 22 years for HbAA and HbSS subjects, accounting for the potentially predictive effects of age and genotype as a linear regression, and for multiple measurements per subject. This model included an interaction term (age * genotype) to allow for different rates of spleen length change in each genotype and is used to formally assess the size of genotype difference in spleen length at ages 12, 14, 16, 18, 20, and 22 years.

Results

Failure to visualize the spleen

There were marked differences between patients and controls in the ability to visualize the spleen by ultrasound. Of the 128 scans in HbAA controls, all were measurable whereas this applied to only 1103/2,138 (52%) in HbSS (Table 1). Non-visualized spleens in HbSS increased with age from 34% in those aged 6.0–7.9 years to 72% in those aged 24 years and above. Three patterns were discernable, the spleen was never visualized in 568 scans in 56 patients, was not visualized on occasional visits in 134 scans in 15 patients, and splenic length progressively declined in 216 scans in 32 patients.

Table 1.

Spleens visualized/not visualized on ultrasound by age and gender

Genotype		Males					Females
	Age group (y)	Subjects	Exams	Visualized	Not visualized		Subjects	Exams	Visualized	Not visualized
		n	n		n	(%)	n	n		n	(%)
HbAA
	14.0–15.9	-	-	-	-		5	5	5	0	(0)
	16.0–17.9	19	20	20	0	(0)	23	25	25	0	(0)
	18.0–19.9	18	21	21	0	(0)	27	30	30	0	(0)
	20.0–21.9	3	3	3	0	(0)	9	11	11	0	(0)
	22.0–23.9	2	2	2	0	(0)	5	6	6	0	(0)
	24+	3	3	3	0	(0)	2	2	2	0	(0)
Total		45	49	49	0	(0)	71	79	79	0	(0)
HbSS
	6.0–7.9	18	24	18	6	(25)	16	23	13	10	(43)
	8.0–9.9	39	62	37	25	(40)	34	67	43	24	(36)
	10.0–11.9	56	113	73	40	(35)	48	102	64	38	(37)
	12.0–13.9	63	156	82	74	(47)	69	140	85	55	(39)
	14.0–15.9	76	175	93	82	(47)	57	157	88	69	(44)
	16.0–17.9	68	173	102	71	(41)	62	156	76	80	(51)
	18.0–19.9	60	154	80	74	(48)	55	142	65	177	(54)
	20.0–21.0	55	120	46	74	(63)	43	104	49	155	(53)
	22.0–23.9	41	85	28	57	(67)	36	76	30	46	(61)
	24+	23	55	10	45	(82)	26	55	21	34	(62)
Total		499	1117	569	548		446	1022	534	488

HbSS, homozygous sickle cell disease.

Numerical values in measured spleens

Splenic lengths, unadjusted for height in HbAA and HbSS genotypes (Table 2) confirm lower values in HbSS (mean 77–103 mm in males, 70–83 mm in females) compared with normal controls (mean 89–101 mm males; 86–95 mm females), these differences persisting after modelling corrected for the known effects of gender and body height (Table 3). This modelling in HbSS demonstrated a steeper age-related decline (1 year increments in age showing an adjusted spleen size of −1.42 mm/m/year, p < 0.001) compared to controls (−1.11 mm/m/year, p < 0.001).

Table 2.

Unadjusted splenic lengths by age and gender in HbAA, and HbSS

Genotype		Males			Females
	Age group (y)	n	Mean, SD	Median (5, 95%)	n	Mean, SD	Median, (5, 95%)
HbAA
	14.0–15.9	-	-	-	5	91, 7	94 (82, 98)
	16.0–17.9	19	99, 12	100 (76, 121)	25	91, 8	89 (81, 106)
	18.0–19.9	18	101, 10	102 (86, 119)	30	93, 13	95 (72, 115)
	20.0–21.9	3	93, 6	96 (87, 97)	11	90, 13	89 (71, 107)
	22.0–23.9	2	89, 11	89 (81, 97)	6	86, 8	88 (77, 97)
	24+	3	93, 14	94 (79, 106)	2	95, 4	95 (92, 98)
Total		45			71
HbSS
	6.0–7.9	18	77, 14	80 (55, 109)	16	70, 13	68 (49, 94)
	8.0–9.9	39	77, 15	75 (59, 103)	34	74, 19	73 (50, 105)
	10.0–11.9	56	80, 19	77 (55, 110)	48	77, 25	72 (47, 119)
	12.0–13.9	63	82, 24	83 (44, 125)	69	80, 30	74 (46, 130)
	14.0–15.9	76	85, 28	83 (42, 140)	57	80, 30	74 (44, 147)
	16.0–17.9	68	81, 25	78 (48, 133)	62	78, 23	76 (46, 121)
	18.0–19.9	60	83, 28	77 (48, 141)	55	89, 24	76 (52, 131)
	20.0–21.0	55	84, 28	84 (43, 130)	43	78, 23	74 (51, 129)
	22.0–23.9	41	90, 30	82 (59, 176)	36	81, 27	72 (52, 135)
	24+	23	103, 43	89 (50, 173)	26	83, 26	73 (56, 118)
Total		499			446

HbSS, homozygous sickle cell disease.

Table 3.

Modelled splenic length adjusted for body height and gender by hemoglobin genotype

	Hemoglobin genotypes		Difference p value
Age (y)	HbAA	HbSS	AA-SS	Significance
	Mean (5th–95th )	Mean (5th–95th )
12.0	68.4 (59.1, 77.7)	55.5 (52.6, 58.3)	12.9	0.01
14.0	65.5 (58.7, 72.4)	52.7 (50.0, 55.4)	12.9	0.001
16.0	62.7 (58.0, 67.3)	49.9 (47.2, 52.5)	12.8	<0.001
18.0	59.8 (56.3, 63.3)	47.0 (44.3, 49.8)	12.8	<0.001
20.0	56.9 (52.6, 61.2)	44.2 (41.4, 47.1)	12.7	<0.001
22.0	54.0 (47.7, 60.4)	41.4 (38.3, 44.5)	12.6	<0.001

HbSS, homozygous sickle cell disease.

Patients with HbSS had multiple sequential observations providing both cross-sectional and longitudinal data but regression analysis confined to subjects with 2–13 annual observations did not materially change the results. The effects of correction for body height is demonstrated dramatically in Figure 1, when a gradual increase in splenic length with age in HbSS, changes to an age-related decline.

Figure 1.

Median spleen length (dotted line) with 25th and 75th percentiles (shaded areas) in controls (HbAA) and patients (HbSS) in males and females, before (A) and after (B) adjustment for body height. HbSS, homozygous sickle cell disease.

The effect of genetic factors influencing sickling

Modelling in HbSS indicated that adjusted spleen size increased by 2.21 mm/m/year for each 1% increase in HbF and that the spleen size in homozygous α thalassemia increased by 23.46 mm/m/year compared to a normal α globin gene complement) (Table 4). In particular, the effect of homozygous α thalassemia (α−/α−) was highly significant whereas the difference between heterozygous α thalassemia (α−/αα) and normal (αα/αα) was not significant.

Table 4.

Estimated effects of age, HbF and α thalassemia on adjusted spleen size (spleen length/body length) in HbSS.

Factor	Unit change	Spleen size mm/m/yearEffect (95% CI)	Significance
Age	Year	−1.42 (-1.64,–1.20)	p < 0.001
HbF	1% increase	2.21 (1.49, 2.93)	p < 0.001
α thalassemiaa
	11 vs 22	23.46 (11.03, 35.89)	p < 0.001
	12 vs 12	3.24 (-2.35, 8.83)	p = 0.26

HbSS, homozygous sickle cell disease.

Estimations based on 1394 observations in 190 patients.

^a11 (homozygous alpha thalassemia a-/a-)

^b12 (heterozygous alpha thalassemia a-/aa)

^c22 (normal aa/aa).

Clinical splenomegaly

The relationship of unadjusted splenic length to clinical splenomegaly in HbSS subjects aged 18–26 years (Table 5) shows clinical splenomegaly in 10/69 (14%) although the prevalence rose to 50% in those with splenic lengths ≥ 150 mm.

Table 5.

Unadjusted spleen length by ultrasound (≥ 100 mm) and clinical splenomegaly in HbSS aged 18–26 years

Spleen size (mm)	Number of observations	Clinical splenomegaly
100–109	25	2 (8%)
110–119	15	0
120–129	5	0
130–139	7	1 (14%)
140–149	5	1 (20%)
150–159	4	2 (50%)
≥ 160	8	4 (50%)
Total	69

HbSS, homozygous sickle cell disease.

Discussion

Normal adult populations without clinical splenomegaly have splenic volumes between 178ml ¹⁴ and 237 ml ¹⁵ , estimates usually based on CT or scintigraphy which are expensive, time consuming and expose the patients to radiation. Splenic length, readily measured by ultrasound, is closely associated with splenic volume on CT. ^10,15

The median splenic length is influenced by body height and gender ¹⁶ and was 109 mm in healthy Caucasian adults, and 128 mm in adult male autopsies. ¹⁷ Cross-sectional studies show that spleen length increases with age between birth and 18 years, ^18–21 and stabilized or gradually decreased in adults. ^13,21,22 Most data are derived from Caucasian populations, and although the Jamaican population is almost entirely of West African origin, there is no convincing evidence of racial differences between Nigerian, ²³ Indian/Sri Lankan, ^21,24 Jordanian ²⁵ or Chinese populations¹⁹.

Only limited data are available on splenic length in patients with sickle cell disease. A multicenter study of chronically transfused patients aged 5–19 years with HbSS in the USA ²⁶ noted that unadjusted splenic length increased with age from mean values of 82.5–91.4 mm but did not differ significantly from normal values. However, this was an atypical group of subjects with preceding stroke, chronic transfusion and 23/148 (16%) had had splenectomy. Other studies in Turkey ²⁷ and Greece ²⁸ were essentially qualitative rather than quantitative.

Clinical observations suggest that the spleen in HbSS initially enlarges followed in most patients by a progressive fibrosis and loss of function ¹ but this overall pattern may be modified by episodes of enlargement trapping a significant proportion of red cells either suddenly in acute splenic sequestration (ASS) or chronically in sustained hypersplenism. Splenic palpation is routine clinical care of patients with sickle cell disease but subject to observer variation, posterior position of the spleen and the fact that clinical splenomegaly does not always correlate with splenic volume. ^29,30 Clinically significant ASS has been reported in HbSC with a spleen that was not clinically palpable. ³¹ Recognizing the relative inaccuracy of splenic palpation and the fact that the spleen may be 2–3 times its normal volume before it can be palpated, a non-invasive method of assessing splenic volume may be of value in excluding a possible role in subclinical red cell sequestration. The clinical relevance is that splenic enlargement detectable by ultrasound but not evident from clinical splenomegaly may be useful in the appraisal of ASS where repeated clinical events may be an indication for splenectomy. ⁵

Body height is an important associate of splenic length and the slowed physical growth in HbSS ³² confounds interpretation of splenic length but adjustment for body height creates a clearer picture (Figure 1), which confirms an age-related reduction in splenic length consistent with the splenic fibrosis of HbSS. In this context, and considering the strong influence of body height on spleen length, a ratio of spleen length/body height would be preferable.

Furthermore, the effects of genetic factors known to inhibit sickling such as α thalassemia and HbF level could also be studied. Heterozygous α thalassemia (α−/αα) occurs in 34% and homozygous α thalassemia in 3% of the Jamaican population. ¹³ Although α thalassemia may be uncommon in indigenous populations of the UK and USA, it occurs in 30% of West African and 10% of Asian immigrants and so is an increasingly important issue in the UK where data indicate immigrants account for 14% of the overall population and 37% of the population in London. ³³ Another genetic factor inhibiting sickling is HbF levels falling from 60 to 80% at birth but the fall is much slower in HbSS where by the age of 5 years, the mean HbF level of 8.3% in HbSS ³⁴ compared with 0.6% in controls. HbF inhibits polymerization of deoxygenated HbS and high levels of HbF are known to inhibit sickling and frequently ameliorate the clinical course of the disease. Both factors have been shown in the present study to reduce the age-related decline in splenic length.

Limitations of the study include the availability of a consultant radiologist for only 3 weeks annually, the rural and often distant residence of many of the patients, the lack of data in younger HbSS patients prior to the 1988 start of the study, the time constraints which limited tests on the controls, and the less complete follow-up of controls invited to attend during the 1992/1993 window. Furthermore, the loss of HbSS subjects by death and emigration may have introduced biases into the studied population.

The ages of subjects in the current study (6–24 years) are older than that for acute splenic sequestration which is uncommon after 5 years but chronic hypersplenism does affect this age group. The well recognized inaccuracy of clinical splenic palpation as an indicator of spleen size is confirmed yet again and it is possible that red cell sequestration occurs within a spleen that is not palpable and so splenic ultrasound may be a useful adjunct to clinical management.