Newborn bloodspot results: predictive value of screen positive test for thalassaemia major

Abstract

Aim

There are limited published data on the performance of the percentage of haemoglobin A (Hb A) as a screening test for beta thalassaemia major in the newborn period. This paper aims to analyse data derived from a national newborn bloodspot screening programme for sickle cell disease on the performance of haemoglobin A (Hb A) as a screening test for beta thalassaemia major in the newborn period.

Methods

Newborn bloodspot sickle cell screening data from 2,288,008 babies were analysed. Data reported to the NHS Sickle Cell and Thalassaemia Screening Programme in England for the period 2005 to 2012 were also reviewed to identify any missed cases (4,599,849 babies).

Results

Within the cohort of 2,288,008 births, 170 babies were identified as screen positive for beta thalassaemia major using a cut-point of 1.5% HbA. There were 51 identified through look-back methods and 119 prospectively identified from 4 screening laboratories. Among 119 babies with prospective data, 7 were lost to follow up and 15 were false positive results. Using a cut-off value of 1.5% Hb A as a percentage of the total haemoglobin as a screening test for beta thalassaemia major in the newborn provides an estimated sensitivity of 99% (from the look back arm of the study) with a positive predictive value of 87% (from the prospective arm of the study). Excluding infants born before 32 weeks gestation, the positive predictive value rose to 95%.

Conclusion

A haemoglobin A value of less than 1.5% is a reliable screening test for beta thalassaemia major in the newborn period.

Introduction

Beta thalassaemias are a group of hereditary disorders of the blood that result in a reduced or absent production of Hb A, which is an essential component of the main adult haemoglobin, Hb A. Methods of newborn screening for sickle cell disease typically examine the proportions of Haemoglobins F, A and variant haemoglobin eluted from dried bloodspots. A decreased Hb A pattern will routinely be seen in such a process, and this is predictive of beta thalassaemia major in normal gestation newborns.

Beta thalassaemia is usually caused by point mutations and, less frequently, deletions of the beta globin gene located on chromosome 11, leading to either reduced (beta⁺) or absent (beta⁰) synthesis of the beta globin chains. Inheritance is predominantly autosomal recessive, although some dominant mutations have been reported. Mutation type and inheritance patterns result in variable phenotypes, which range from severe anaemia to clinically asymptomatic carriers, differentiated into three types, termed thalassaemia major, thalassaemia intermedia, and the carrier state, thalassaemia minor. ¹

Prior to the introduction of red cell transfusion therapy, presentation with severe anaemia in the first two years of life, followed by death, usually by the age of five, was the typical course for a child with beta thalassaemia major. Today, treatment is via lifelong transfusion therapy, which has associated complications of iron overload, including endocrine dysfunction, cardiomyopathy, liver fibrosis, and cirrhosis. ¹ A demanding daily regime of therapy with iron chelators, which enable iron excretion from the body, is required to treat these complications. Insufficient red cell replacement results in symptoms of anaemia, jaundice, poor growth and muscular development, masses from extramedullary haemopoiesis, and skeletal changes from expansion of the bone marrow. ¹

Thalassaemia intermedia is a highly variable condition. Although some cases are transfusion-dependent, it is usually a less severe disease than beta thalassaemia major.

In the United Kingdom (UK), the National Screening Committee (NSC) recommended the introduction of screening for sickle cell disease in the newborn in 2002, but did not support the case for screening for thalassaemia in the newborn. This meant that when the NHS Sickle Cell and Thalassaemia Screening programme (NHS SC&TP) was established, a decision had to be made on reporting incidental findings suggesting beta thalassaemia major, despite the lack of formal support for a screening programme. ² Since 2003, the NHS SC&TP has recommended using a cut-off value of less than 1.5% for such incidental reporting of cases, and since 2005 the programme has collected England-wide data on these cases. In the absence of published data, these cut-off values were derived using expert opinion, information on the variation of Hb A by gestation, and existing practice, and allowed for the lack of sensitivity and expected instrument background noise at this percentage of Hb A.

In 2011 the NHS SC&T commissioned a formal review of evidence, which was again considered by the UK NSC. Newborn screening for thalassaemia was not supported by the NSC, although it was agreed that the policy of reporting clinically significant thalassaemias, found incidentally in the course of newborn screening for sickle cell disease, should continue (to avoid moribund infants presenting acutely) until more evidence becomes available. ³ The NSC concluded that “there is uncertainty about whether the current test cut-off value is appropriate for the population as a whole”. ⁴ They asked that the NHS SC&T Screening Programme work with clinicians and laboratories to facilitate rigorous evaluation to obtain new knowledge regarding the working of the relevant cut-offs and to facilitate answering “uncertainties regarding treatment due to the lack of evidence to support treatment before symptomatic presentation”, which were the two reasons given for not supporting screening. ⁴ We here provide nationally derived information on the appropriate cut-off value for such a screening programme in England to identify babies with beta thalassaemia major.

Methods

The data presented here are based on the results of laboratory screening in England, using the newborn bloodspot card collected as part of the newborn bloodspot screening test. The test is offered to all babies in England, and is designed to detect five conditions, including sickle cell disease. In England, a sample is taken from babies at age 5--8 days. All abnormal results from the first screen, including low percentages of Hb A, are referred for confirmation by a second line test, using a different analytical principle, before the screening results are reported. It is these twice tested screen positive results that are reported here.

The NHS SC&T Screening Programme began to collect data on disease status and incidence of haemoglobinopathies from all 13 newborn laboratories in 2005. Therefore, a complete overview dataset was not available from all areas for the entire time period of this study.

The data presented here are based on three analyses:

Group One – all babies screened (n = 4,599,849) in a 7 year period from 2005 to 2012 (Table 1). Four laboratories reported the majority of beta thalassaemia disease cases. These were used for a more detailed study (Groups 2 and 3)

Group Two, a sub-set of Group One (samples tested at Central Middlesex, Leeds, Manchester, n = 1,515,086). These laboratories provided data on all samples tested, with an Hb A percentage of less than 1.5 (prospective data arm of study, data period 2004 to 2012).

Group Three (samples screened at Birmingham, n = 772,922) This laboratory carried out a look-back exercise (data period 2001 to 2012) on confirmed beta thalassaemia babies being treated in their paediatric centre. Specific Hb A percentages were not available from this centre, but the values were all less than 1.5%.

Table 1.

Total number of suspected beta thalassaemia disease cases reported over period 2005--2012.

Laboratory catchment	2005--2012
	Hb A <1.5%	Total screened	rate/ 100,000
Bristol	0	284,893	0.00
Cambridge	0	190,657	0.00
C Middlesex & GOSH	47	826,169	5.69
Leeds	23	298,249	7.71
Liverpool	1	202,498	0.49
Manchester	29	390,668	7.42
Newcastle	1	243,077	0.41
Oxford	4	175,312	2.28
Portsmouth	2	229,091	0.87
Sheffield	11	510,737	2.15
South East Thames	3	401,862	0.75
South West Thames	6	354,776	1.69
West Midlands/Birmingham	35	491,860	7.12
England total	162	4,599,849	3.52

For the more detailed analysis in Groups 2 and 3, a review of newborn bloodspot screening results defined as requiring follow up (screen positive) for beta thalassaemia was undertaken. Data were requested for all babies referred for clinical follow-up because they had Hb A percentages of less than 1.5 on the preliminary screening test. Data were collected on initial result, gestational age, birth weight, sex, parental screening results, and final diagnosis, though not all laboratories could provide all these data on every baby.

In addition, all cases of beta thalassaemia with a clinical presentation reported to the programme as not having been detected by the newborn screening programme in England were reviewed.

Results

Table 1 shows that, for the period where data were reported to the NHS SC&T (2005 to 2012), the suspected beta thalassaemia disease cases totalled 162, and that the four laboratories in the detailed study (Central Middlesex/Great Ormond Street (GOSH), Leeds, Manchester, and Birmingham/West Midlands) contributed the majority of screen positive results in England (83%). The birth prevalence of screen positive results nationally was 3.52 per 100,000 (1:28,400). This ranged from areas with only one case or no cases over the period, to Leeds, with a prevalence of 7.71 per 100,000, Manchester with 7.42 per 100,000 and West Midlands with 7.12 per 100,000. These relatively high rates are attributed to larger Pakistani and south Asian populations which are mainly concentrated in these areas, and the larger Bengali and Indian populations which are mostly located in north London (the other area with high prevalence).

The screening results from 2,288,008 babies were used for this survey, and represent the total denominator for estimating false negative results.

The Group Two analysis (prospective data arm) was for the period 2004 to 2012, and covered London, which has large African, Caribbean, Asian, and Bengali populations. There was a total of 1,515,086 babies in this arm of the study; 119 babies were identified as having potential beta thalassaemia major or intermedia (reported as suspected beta thalassaemia disease); of these seven were lost to follow up by the laboratory, including three with low birthweight who had died. There remained 112 screen positive results with follow-up, which were subsequently confirmed as: beta thalassaemia major (90), beta thalassaemia intermedia (7), and false positive (15). The breakdown of the cases among the three laboratories is shown in Table 2.

Table 2.

Diagnostic results of babies with screen positive results.

True positives	False positives	Total
Central Middlesex	48	11	59
Leeds	24	2	26
Manchester	25	2	27

These results give a positive predictive value of 86.6% (95% CI, 78.9 to 92.3) for thalassaemia major and Intermedia combined (97/112) with the cut-off of Hb A less than 1.5%, using the methods recommended by the screening programme.

The Group Three (look-back arm) covered the period 2001 to 2012, with a total of 772,922 babies; 51 of these were identified with Beta thalassaemia major (49) or intermedia (2). All 51 babies were reported to have had screen positive results on initial bloodspot screen, and no false negatives were reported. Information on false positives was not complete, so has not been used to calculate the positive predictive value. These data suggest that population sensitivity for this area, with a predominantly Pakistani Asian population, and more expertise in beta thalassaemia, was 100% (although specificity and predictive value cannot be ascertained).

False positives

From Group Two, of the 15 false positive screens, two had no Hb A (both had Hereditary Persistence of Fetal Haemoglobin [HPFH] with a carrier state for a haemoglobin variant or thalassaemia). Eleven were premature babies of less than 32 weeks’ gestation (range 25 to 29 weeks). For comparison, of the 97 true positive cases, 67 had a gestational age recorded, and there was only one case with a gestational age of less than 32 weeks. The remaining 4 false positives were over 32 weeks gestation, and of these one was a beta thalassaemia carrier and one a beta thalassaemia carrier with HPFH. Details of the false positive results are given in Table 3.

Table 3.

Outcome on babies with false positive screen result.

Outcome	Number	Gestation (weeks)	Mother
No abnormality	7	25--29
No abnormality	3	34--38	B thalassaemia  carrier (1 case)
Hb AS	3	27--29
HPFH with variant	1	30
HPFH with beta  thalassaemia carrier	1	41	B thalassaemia  carrier

False negatives

There were no missed cases reported by the four laboratories taking part in this survey (covering approximately 2.5 million births,) in either the prospective or the look-back arms. However, using the 1.5% cut-off in isolation would have resulted in one false negative result. This case had an Hb A percentage of 1.9, but was reported as screen positive by the investigating laboratory. The case was known to be at risk from the parental results, and subsequent DNA analysis confirmed that the baby was homozygous for the IVS1-110 (G > A) beta⁺ mutation which, in the homozygote, invariably results in a transfusion dependent condition. Without the parental information, this case would have been missed by the Hb A cut-off of 1.5%, and is counted as a false negative for the purposes of data analysis, even though the case was referred for clinical follow-up and not missed. If this case is included, there is a false negative rate of (1/ 1,515,086 and a specificity of 99.9% (1,514,973/1,514,973 + 15). This means that, without additional information that might be available from antenatal screening, 1 of 97 cases of thalassaemia major or intermedia would be missed by the specified cut-off, yielding a sensitivity of approximately 99% (1/97 cases missed).

To summarize the Group Two findings, of the 112 babies with a confirmed result, 97 were beta thalassaemia major or intermedia (86.6%), and two had no Hb A for other reasons. Of the remaining 13 remaining babies, only three were over 32 weeks’ gestation. This demonstrates a high specificity and positive predictive value (95%) for the test in newborns over 32 weeks’ gestation, and a need for caution in interpreting results for those babies born before 32 weeks. This is consistent with the gradual increase in the production of adult haemoglobin in the newborn.

Incident reporting of missed cases (possible unscreened cases or false negatives)

Three cases were identified through incident reporting. All fell outside of the four study laboratories catchment areas, and all were a result of co-inheritance of a beta⁺ and a beta⁰ mutation. These are shown below.

These cases were observed over approximately five years and include 3,000,000 screens. During this time, between 19 and 29 cases of suspected beta thalassaemia major were reported to the programme each year. The data support the findings of the main analysis, which suggests that clinically significant cases of beta thalassaemia major are unlikely to be missed using the current approach. They also suggest that the sensitivity in detecting cases where there is some production of Hb A at birth may be lower than in those cases where there is none.

In summary, the combined findings from the routine reporting to the programme, more detailed information from the four laboratories with largest numbers of cases (and most expertise in these conditions), and incident reporting to the programme indicate that the policy of reporting of incidental findings of Hb A of less than 1.5% works very effectively as a screening test. These findings also show that incident reporting centrally to the programmes provides information which can inform policy, and has shown that all undetected cases were the result of co-inheritance of a beta⁺ and a beta⁰ mutation.

Discussion

This study suggests that a cut-off value of 1.5% Hb A provides a reliable screening test to identify cases at risk of beta thalassaemia major. The positive predictive value, specificity, and sensitivity are good, and higher than other recently reported results for Medium-chain acyl-CoA dehydrogenase deficiency (MCADD), which were reported to be 77% based on a similar number of cases. ⁵ They demonstrate that the test performance is not a significant barrier to the recommendation of screening for thalassaemia major in the newborn period.

The limitations of the study are that it is not a complete country report, that numbers of positive cases are only about 100, and that we did not have complete data on all false positives. We have reviewed all incidents reported to the programme, and found that cases involving beta plus mutations have been missed by the action value of 1.5%, and that three of the four cases reported are known to be transfusion dependent. However, as a screening programme, this still provides a high level of detection. This study does not review the effect of changing the cut-off value from 1.5%. Data were not available in this study on the percentage of Hb A in all cases. The study could detect only whether the Hb A was less than 1.5%. A three year study is currently being undertaken to review the percentage of Hb A, in conjunction with genotype results. Those babies subsequently confirmed to have beta thalassaemia major/intermedia without prenatal diagnosis or DNA data available will require DNA testing to confirm the beta gene mutations for correlation with the Hb A% value. These data will allow the cut-off to be set based on evidence, rather than on expert opinion.

The data in this paper demonstrate that, for babies born at over 32 weeks’ gestation, using current cut-offs, the positive predictive value is high, and at least comparable or better than the positive predictive value in other recently introduced newborn screening tests. ⁵ The lower reliability of the cut-off under 32 weeks’ gestation is consistent with other reports, which show the challenges of interpreting haemoglobinopathy results in premature babies. ⁶ In the normal fetus, the main haemoglobin is fetal haemoglobin (Hb F), but in late fetal life a switch of haemoglobin synthesis occurs, and Hb F is gradually replaced by Hb A. The amount of Hb A at birth can be highly variable. In a term baby it is usual to observe Hb A levels between 7.5 and 25% of total haemoglobin. In a premature baby this percentage is reduced proportionally with the degree of prematurity as shown in Figure 1. ² This is recognized in practice, with guidance that results on premature babies should be interpreted with caution. ²

Figure 1.

Hb A% in routine specimens taken from 30,000 untransfused babies with gestations from 23 to 42 weeks.

Conclusion

These data show that using measurement of Hb A as a percentage of total haemoglobin as a screening test for thalassaemia major in the newborn period provides reasonable sensitivity, specificity, and positive predictive value. These data provide support for an Hb A cut-off of 1.5% as a screening test (if all other screening criteria are met), for thalassaemia major.