Fetal haemoglobin levels and haematological characteristics of compound heterozygotes for haemoglobin S and deletional hereditary persistence of fetal haemoglobin


Dr D. Ngo, Department of Medicine, Boston University Medical Center, 820 Harrison Ave, FGH 1025, Boston, MA 02118, USA.
E-Mail: duyen.ngo@bmc.org


Compound heterozygotes for sickle haemoglobin (HbS) and hereditary persistence of fetal haemoglobin (HPFH) have high fetal haemoglobin (HbF) levels but few, if any, sickle cell disease-related complications. We studied 30 cases of HbS-HPFH (types 1 and 2), confirmed by molecular analysis, and report the haematological features and change in HbF levels over time. These results were compared to those of patients with sickle cell anaemia or HbS-β0 thalassaemia, including a subgroup of patients carrying the XmnI polymorphism, known to be associated with elevated HbF. Among the HbS-HPFH patients, HbF level was 50–90% during infancy and declined steeply within the first few years of life, stabilizing between ages 3 and 5 years, at approximately 30%. Mean HbF of individuals age 5 or older was 31 ± 3%, average haemoglobin concentration was 130 ± 10 g/l and average mean corpuscular volume (MCV) was 75 ± 4 fl. Univariate and multivariate regression analyses significantly associated HbF with age, haemoglobin concentration, and MCV (< 0·001). There was a strong inverse association between HbF and age (r = −0·9, P < 0·001). Despite having a much higher HbF level, patients with HbS-HPFH have a similar age-related pattern of HbF decline and associations as patients with sickle cell anaemia or HbS-β0 thalassaemia.

Elevated fetal haemoglobin (HbF) levels ameliorate some clinical features of sickle cell disease by reducing HbS content and retarding HbS polymerization (Akinsheye et al, 2011). Hereditary persistence of fetal haemoglobin (HPFH) characterizes a group of phenotypically and genetically heterogeneous conditions marked by a substantially elevated HbF level in adulthood. HPFH is divided into two main groups: deletional and nondeletional. Deletional HPFH is caused by variable length deletions (13–106 kilobases) in the β-globin gene (HBB) cluster leading to decreased or absent β-globin synthesis and variable compensatory increases in γ-globin synthesis with a pancellular or homogenous distribution of HbF among erythrocytes. Nondeletional HPFH encompasses a broad category of disorders with elevated HbF, typically distributed heterocellularly, resulting from mutations in the γ-globin gene (HBG2, HBG1) promoter regions or inheritance of HbF modulating quantitative trait loci (QTL), which may be unlinked to the HBB cluster (Forget, 1998; Akinsheye et al, 2011).

The two most common types of deletional HPFH in people of African descent (types 1 and 2) are characterized by >80 kb deletions (Fig 1) involving HBD (δ-globin gene) and HBB that are staggered by approximately 5 kb at the 5′ and 3′ ends (Forget, 1998). Heterozygotes for HPFH 1 and 2 have similar HbF levels (20–28%) and mean corpuscular volume (MCV) but differ in the ratio of Gγ to Aγ chains (approximately 50:50 in type 1 and 30:70 in type 2) (Kutlar et al, 1984; Bakioglu et al, 1986). Homozygotes for HPFH type 1 and 2 are clinically unaffected and have 100% HbF and mild microcytosis. Prior to the advent of molecular techniques, a survey of 10 000 African Americans in Baltimore found the rate of HPFH carrier state to be 0·2% (Charache & Conley, 1969). Patients who were compound heterozygous for HbS and HPFH have a relatively benign clinical course with nearly normal haematological parameters and HbF levels between 15% and 35% (Conley et al, 1963; Charache & Conley, 1969).

Figure 1.

 Relative positions of HPFH 1 and 2 deletions. HPFH 1 deletion results in an 85 kb deletion on chromosome 11: 5 174 451 to 5 259 368. HPFH 2 deletion results in an 84 kb deletion on chromosome 11: 5 179 688 to 5 263 979. Coordinates are based on the human genome reference sequence 19. Arrows represent the hypersensitive sites of the HBB control region (β-LCR). HPFH, Hereditary persistence of fetal haemoglobin.

In prior series, confirmation of the exact mutations was not available and genetic inheritance was deduced from family evaluation. We studied 30 individuals with HbS-HPFH type 1 or 2 in whom the HPFH deletion was confirmed by DNA-based diagnostics and report the analyses of haematological parameters, clinical features, and longitudinal HbF data. We also compare the pattern of HbF decline in these cases with other genotypes of sickle cell disease to better understand this pattern in the different groups.


The records of the Haemoglobin Diagnostic Reference Laboratory at Boston Medical Center were used to identify cases of sickle cell anaemia with elevated HbF (>10%) referred for haemoglobin diagnostic testing between 2004 and 2010. The sickle cell mutation was confirmed by the presence of the glutamine to valine mutation at position 6 of the HBB gene. We identified 30 cases of compound heterozygotes for HbS and deletional HPFH (types 1 or 2) confirmed by DNA-based multiplex gap-polymerase chain reaction tests designed specifically to detect HPFH 1 and 2 mutations. A total of 72 HbF values were collected for these subjects at various ages.

Laboratory and clinical data, including age, gender, complete blood count, and haemoglobin high performance liquid chromatography (HPLC) results, were obtained from information provided to the laboratory at the time of haemoglobin diagnostic testing. Additional data regarding longitudinal HbF levels, complete blood counts, and clinical events were obtained via chart review at the participating institutions with approval by their local institutional review boards. Data regarding age at which HbF was measured enabled us to study the change in HbF over time. For comparison purposes, we obtained demographic (e.g. age, gender) and laboratory data, including HbF measurements, collected through the Cooperative Study of Sickle Cell Disease (CSSCD) for 4085 patients with sickle cell anaemia or HbS-β0 thalassaemia. For intergroup comparison, a subset of patients who had at least one confirmed XmnI polymorphic site (C>T at nucleotide −158 bp upstream of the HBG2 gene), known to be associated with elevated HbF, was segregated from the rest of the CSSCD population. Thus, two genetically-defined groups (HbS-HPFH and homozygous HbS with the XmnI polymorphism) were available for comparison with the general CSSCD population.

Continuous values for age, haemoglobin, MCV, and HbF for the HbS-HPFH group and XmnI carriers were compared to the CSSCD population using the t-test. The relationship between HbF percentage and age was examined visually within each group by scatterplot. Loess curves were generated to allow for inter-group comparisons. Given the observed pattern of HbF decay, age was natural log-transformed for all subsequent analyses. Due to the skewed distribution of HbF values, Spearman correlation was used to assess the relationship between HbF and log-transformed age. Univariate and multivariate analyses were used to assess the relationship between HbF, gender, type of HPFH (1 or 2), haemoglobin level and MCV. All analyses were performed using Stata/SE11.1 (Stata Corp, College Station, TX). P values < 0·05 were considered statistically significant.


Seven cases of HPFH type 1 and 23 cases of HPFH type 2 were studied, 12 were males. Although HPFH type 1 has been reported to be the most common deletional HPFH occurring in American blacks, our sample of patients contained more HPFH type 2 than type 1, which may be due to referral bias (Collins et al, 1987). There were 72 HbF values from the newborn period to age 20 years with an average of 40·6% (median 35·4%, range 27·1–90·8%). There was no significant difference between the two types of HPFH regarding HbF, total haemoglobin, and MCV (Table I).

Table I.   Comparison of haematological values for HbS-HPFH types 1 or 2.
 HPFH 1 (n = 7, 2 males)HPFH 2 (n = 23, 10 males)t-test P-value*
Mean (SD)RangeMean (SD)Range
  1. HPFH, Hereditary persistence of fetal haemoglobin; HbS-HPFH, compound heterozygotes for HbS and HPFH; MCV, mean corpuscular volume; SD, standard deviation.

  2. *There was no significant difference between the average HbF (in patients ≥5 years old), haemoglobin, or MCV between these groups based on t-test analysis.

Average HbF (%)32·6 (3·0)28·7–36·031·0 (2·3)27·1–35·20·26
Haemoglobin (g/l)130 (8·0)119–149125 (12)100–1480·07
MCV (fl)73·7 (6·2)65·0–93·674·9 (6·4)65·0–99·20·48

Haematological values in HbS-HPFH and other study groups are shown in Table II. For patients in the HbS-HPFH group, the mean haemoglobin concentration was 127 ± 11 g/l, average MCV was 74·6 ± 6·3 fl, and average HbF for those ≥5 years of age was 31·3 ± 2·4%. Two hundred and seventy-eight CSSCD patients with sickle cell anaemia had at least one copy of the XmnI polymorphism (rs7482144). In XmnI carriers, the average HbF for those ≥5 years of age was 7·6 ± 4·4%, mean haemoglobin concentration was 90 ± 15 g/l with an average MCV of 88·0 ± 10·3 fl.

Table II.   Mean haemoglobin, mean corpuscular volume (MCV), reticulocyte percentage, and HbF values for 3 groups of sickle cell patients.
Mean (SD)RangeMean (SD)Ranget-test P-value*Mean (SD)Ranget-test P-value*
  1. CSSCD, Cooperative Study of Sickle Cell Disease, a database containing patients with HbSS and HbS-β0 thalassaemia; HbS-HPFH, compound heterozygotes for HbS and HPFH; XmnI, sickle cell patients from the CSSCD with at least one XmnI polymorphism; SD, standard deviation.

  2. *T-test P-value using CSSCD as reference group.

Age (years)13·5 (12·5)0–66·26·1 (6·0)0–20 12·3 (11·8)0–55·0 
Haemoglobin (g/l)93 (18)49–165127 (11)100–149<0·0190 (15)62–140<0·01
MCV (fl)84·7 (10·5)34–12074·6 (6·3)65·0–99·2<0·0188·0 (10·3)55–113<0·01
HbF (%)8·4 (9·3)0·1–68·540·6 (14·7)27·1–90·8<0·0112·4 (11·2)0·6–71·5<0·01
Reticulocytes (%)9·8 (7·2)0–49·01·4 (0·6)0·5–2·8<0·0110·6 (6·0)0·4–31·20·09
HbF (%) when ≥5 years old5·3 (4·2)0·1–35·131·3 (2·4)27·1–36·0<0·017·6 (4·4)0·6–20·1<0·01

In HbS-HPFH subjects, HbF level at birth was 50–90% and declined steeply during the first few years of life, stabilizing between ages 3 and 5 years at approximately 30%. This pattern of decline, taking several years after birth to stabilize, was also present in patients with sickle cell anaemia and HbS-β0 thalassaemia and the subset of patients carrying the XmnI polymorphism. The time for HbF to plateau was similar among the different groups but the level at which HbF stabilized differed, being highest among individuals with HbS-HPFH (Fig 2).

Figure 2.

 Association of HbF and age. (A) compound heterozygotes for HbS and HPFH, (B) sickle cell patients in Cooperative Study of Sickle Cell Disease (CSSCD) database, (C) sickle cell patients who carry at least one XmnI polymorphism, (D) Loess Curves of HbF plots from Figures A, B, C. HPFH, Hereditary persistence of fetal haemoglobin; HbS-HPFH: compound heterozygotes for HbS and HPFH.

A review of the clinical information for HbS-HPFH patients revealed absence of symptoms due to sickle cell anaemia. These patients were typically followed in clinic once a year and did not require any interventions. Penicillin prophylaxis was discontinued after confirmation of HbS-HPFH and they did not have any severe infections. A detailed chart review of two patients with HbS-HPFH type 2 and one patient with HbS-HPFH type 1, aged 6, 15, and 19 years, show that these patients are healthy with no known complications attributable to sickle cell disease. They had no visits to the emergency department or hospitalization for pain or other complications of sickle cell disease.

In univariate and multivariate regression analyses, HbF was significantly associated with age, haemoglobin concentration, and MCV (< 0·001). There was greater than 80% logarithmic linear correlation between HbF and age, plateauing at a HbF level of approximately 30% in adulthood (< 0·001; Fig 3). There was no association between gender and type of HPFH deletion. As a comparison, in sickle cell anaemia or HbS-β0 thalassaemia cases, HbF was also significantly associated with age, haemoglobin concentration, and MCV. An association between HbF and female gender was found in the CSSCD cohort but not in the HbS-HPFH population, perhaps due to sample size.

Figure 3.

 HbF decline with age. There is a strong inverse correlation between HbF and log-transformed age in patients with HbS-HPFH as well as HbSS patients with or without XmnI polymorphism. CSSCD: Cooperative Study of Sickle Cell Disease. HbS-HPFH: compound heterozygotes for HbS and hereditary persistence of fetal haemoglobin.


We report the largest group of molecularly documented HbS-HPFH cases and confirm their nearly normal haematological values and the absence of clinical findings or complications attributed to sickle cell disease. HbS-HPFH patients are subject to a similar age-related HbF decline, as seen in patients with other genotypes of sickle cell disease, despite having a much higher level of terminal HbF. The similar pattern of age-related HbF decline suggests that these groups share a common mechanism for HbF silencing during infancy but a different mechanism for the regulation of terminal HbF.

In normal subjects, HbF production decreases dramatically within the first few months of life and usually achieves a level of less than 2% by 1 year of age (Boyer et al, 1975; Wood et al, 1975). In infants with sickle cell anaemia, the switch from HBG to HBB expression is delayed, resulting in a slower decline in HbF, taking 3–5 years to plateau instead of months as seen in normal infants. Although the level of HbF varies from 2 to 20% among individuals with sickle cell anaemia, it remains constant after the age of 5 years (Adekile & Huisman, 1993). The mechanism accounting for this delayed switching is unknown. One hypothesis suggests that the gradual decline and persistence of HbF may be related to the slower centripetal regression of red (or haematopoietic) marrow to the axial skeleton in the presence of expanded erythropoiesis stimulated by haemolysis (Akinsheye et al, 2011). In HbS-HPFH there is little or no haemolysis, but the protracted switch from HBG to HBB expression is still present, suggesting a non-haemolytic driven process. Given that subjects who are homozygous for deletional HPFH have 100% HbF (Marcus et al, 1997), we can surmize that the HbF decline in HbS-HPFH is related to the presence and influence of the βS chromosome, perhaps due to regulation at the level of the HBG genes.

Fetal haemoglobin prevents sickling by diluting HbS and inhibiting its polymerization. In sickle cell anaemia, the F-cell fraction ranges from 17 to 50% of the red cells and the concentration of HbF per F-cell varies (Charache et al, 1995; Marcus et al, 1997). Not all F-cells contain sufficient concentrations of HbF to prevent sickling. Some patients with sickle cell anaemia, like those with Saudi-Indian or Senegal haplotypes, have unusually high HbF levels, with an average of 11% to 21% (Padmos et al, 1991; Steinberg et al, 1995; Adekile et al, 2007). Although they have a milder clinical course than African American sickle cell patients, the Eastern Saudi sickle cell patients still experience painful episodes, acute chest syndrome, avascular necrosis, and other features of sickle cell disease, albeit less frequently than patients with more usual HbF levels (Padmos et al, 1991). The relatively benign clinical course of those with HbS-HPFH can be attributed to the high intracellular erythrocyte HbF concentration and also to the pancellular distribution, as opposed to the heterocellular pattern, in sickle cell anaemia or HbS-β0 thalassaemia. All erythrocytes in HbS-HPFH patients contain a uniformly distributed HbF in sufficient concentration that affords these patients protection from sickle cell complications despite their high HbS fraction. In solubility experiments, a combined HbF and HbA2 concentration of 30% in the presence of HbS (70%) had an equivalent solubility to a haemoglobin mixture from patients with sickle cell trait containing 60% HbA and 40% HbS (Poillon et al, 1993). Though these experiments do not replicate physiological conditions entirely, the results provide a basis for the observation that HbS-HPFH patients with uniformly distributed HF level around 30% have a benign clinical course similar to those with sickle cell trait who have HbA levels of around 60%.

The terminal HbF and the pancellular nature of HbF distribution is probably the product of the HPFH chromosome (Gibbons et al, 2001). Three possible mechanisms have been proposed to explain the increase in HbF associated with the HPFH deletions: 1) deletion of regulatory sequences affecting HBG expression; 2) juxtaposition of enhancers of HBG normally located downstream of HBB; 3) enhanced interaction of the locus control region (LCR) with HBG due to decreased competition from the deleted HBB (Thein & Wood, 2009).

Given the notably benign clinical course of those with HbS-HPFH and the difficulty in distinguishing HbS-HPFH and sickle cell anaemia with unusually high HbF by HPLC results alone in newborn screening and during infancy, it is worth pursuing molecular testing to confirm the diagnosis of HPFH of those with marked elevated HbF after age 1 year. This would enable appropriate counselling and management considerations, together with an opportunity to allay some of the anxiety associated with the diagnosis of sickle cell disease.

Future studies aimed at understanding regulation and distribution of HbF in HPFH and other patients with unusually high HbF might offer insight into mechanisms that can halt decline of HbF and ameliorate symptoms of sickle cell disease.


This work was supported by: U54 HL070819-07 (DAN), U54 HL 70819 (MHS), R01 HL068970 (MHS), R01 DK069646 (DHKC), and in part by the American Lebanese Syrian Associated Charities (BA, JSH).

Authorship contributions

DAN collected, analysed, and interpreted the data, and wrote the manuscript. BA and JSH contributed data and interpretation of results and wrote the manuscript. IA collected data and reviewed the manuscript. HYL performed the DNA-based diagnostics on these patients. IB assisted in statistical analysis and writing of the manuscript. MHS and DHKC interpreted the data and wrote the manuscript.

Conflict of Interests

The authors have no competing interests.