Blood samples were drawn (once) over a six-month period (March, 2007 – August, 2007) as part of a nation-wide population-screening program for thalassemia and hemoglobin S. This screening center is located in the city of Al Ain, United Arab Emirates (UAE), which is on the border with Oman and in close proximity to Saudi Arabia. This screening program is mandated and fully funded by the government. It is restricted to the citizens of the UAE who intend to marry; therefore, the expatriate population was excluded from this study. In the local UAE population, marriages are arranged, often consanguineous, i.e., between cousins, and cannot proceed without a judge issuing a wedding-license after verifying the screening center report. There were no exclusion criteria, but data was missing in 47 (4.5%) individuals, who could not be used; the remaining 1032 consecutive healthy men and women were included in the final analysis.

The population is strictly tribal, with a total of 67 tribes in the country (as per 1968 population census) 13. The tribe, subtribe, and family group allegiance of 78% of subjects was successfully determined from their last names. Members of at least four of the tribes in this study inhabit territories of other four Gulf countries because until very recently there were no barriers to the movement of people in this region.

A second cohort comprising of seven subjects with BN were from hematology clinic; they were used in analysis of the phenotype and the pattern of inheritance of neutropenia trait. In these subjects, the diagnosis of BN was based on 1) the negative history of use of medications known to cause neutropenia, 2) absence of chronic diseases associated with neutropenia, 3) absence of infection, 4) normal physical examination. Moreover, the laboratory tests confirmed the absence of anemia, thrombocytopenia and anti-nuclear antibodies, normal B₁₂ and folate levels, normal bone marrow examination (in three of seven subjects) and documented neutropenia in at least one family member (in six of seven subjects). In all seven subjects, neutrophil counts were obtained from their medical records. Additionally, in two brothers with multiple ANC < 0.5 × 10⁹ cells/L but without recurrent infections, we excluded a mild form of congenital neutropenia by excluding the presence of the mutations at ELA2 and HAX1 locus 1415.

This study was approved by Al Ain Medical District Human Research Ethics Committee (Protocol # 07/123).

The complete blood count (CBC) was performed once on every individual after the blood was collected in a tube containing EDTA and analyzed using the Cell-Dyn Sapphire (Abbot Diagnostics, USA) analyzer. The hospital laboratory subscribes to an overseas external quality control from the United Kingdom (UK National External Quality Assessment Scheme). The standard deviation index of leukocyte counts varied from -0.77 to 1.33 from the target mean value during the study period. Hence, the laboratory met the analytical standards for CBC.

Data were logged into a computer database and analyzed using SPSS ver.15 for Windows (SPSS Inc., Chicago, Ill., USA). Standard descriptive and analytical statistical methods, such as histograms, scatter plots, linear regression and independent samples t-tests, Spearman's and Pearson's correlation coefficient were used. In addition, the ranges of cell counts were determined by taking the mean ± 2 SD on a Box-Cox (power) transformed scale and back transformation of the normal range limits. The figures were plotted using the SPSS and Excel (Microsoft). The level of statistical significance was 0.05.

There is a paucity of data on neutropenia amongst Arabs. In this study, the prevalence of BN (ANC < 1.5 × 10⁹ cell/L) and marked neutropenia (ANC 0.5 – 1.0 × 10⁹ cell/L) in Arabs is 10.7% and 2.3% respectively. As the subjects with BN periodically have ANC in the normal range (Fig 4 and 5), the observed prevalence of BN in the studied population of Arabs may well be an underestimate. The mean neutrophil count in Arabs is 3.3 × 10⁹/L (range, 0.95–7.6), without any significant difference between the sexes. As ethnicity is known to affect neutrophil count, we compared the mean neutrophil count of the Arab subjects in this study, of mainly young adults, with those previously reported in American Blacks, Whites and Mexicans 2. In Arabs, ANC is lower than the mean neutrophil values earlier reported for younger (age <18 years) and older (age ≥ 18 years) American Mexicans, Whites and older Blacks 2, as is shown by plotting neutrophil counts by age for these ethnic groups (Fig 7).

In this study, the neutrophil count correlated with the monocyte count (r = 0.43). This has not been previously reported and is in contrast to patients with congenital neutropenia who often show a relative increase in monocytes counts. The association of neutropenia and monocytopenia may imply that subjects with BN may also be monocytopenic. This may give a possible clue to its origin. For example, in cyclic congenital neutropenia, synchronized oscillation of neutrophils and monocytes may be related to a back loop inhibition from neutrophils to their common precursor cell 16. Therefore, a concordant decrease of neutrophils and monocytes may suggest that the mechanism may involve a common progenitor cell.

In this study, there was no sex bias as there were equal numbers of healthy males and females. A study of a benign genetic condition in a population that is shortly going to reproduce is most useful for estimating its prevalence. Similarly, there was no selection bias as the blood tests in the study were mandatory without any cost to the subjects; further, we identified 73 different tribes, subtribes and family groups. However, the parameters which may increase neutrophil count, e.g., smoking were not controlled for in the study.

Our findings confirm and extend previous reports on neutropenia in the Arab Peninsula. BN has been previously reported in Jordanians and Bedouin Arabs (from the north of Arabian Peninsula), Kuwaiti citizens (in central part of Peninsula) as well as among Yemeni and Ethiopian Jews (many residing in southern part of the Peninsula), and amongst Arab tribe of Sudanese origin residing in the Middle East 9101112. In addition, four of the tribes with neutropenia in our study are present in four other Arab Gulf countries. This suggests that a high prevalence of BN is likely to be present in most populations in the Middle East and North Africa.

In the second part of the study, to determine the phenotype of BN, we analyzed seven subjects from the hematology clinic. We sorted (in ascending order) all available ANC of these patients, which suggested three possible phenotypes of BN: mild, profound and cycling types of neutropenia. This classification is tentative due to a limited number and irregularity of data points as well as cycling of neutrophil counts in normal subjects and unavoidable background noise 1718. Nonetheless, in several subjects the impressive part was the benign nature of the condition (absence of infections) even when ANC was < 0.5 × 10⁹ cells/L. In one family, two healthy sons had frequently ANC < 0.5 × 10⁹cells/L and a mild form of congenital neutropenia was exclude with the negative tests for mutation at ELA2 and HAX1 gene locus. In contrast to the absence of severe neutropenia in 1032 screened individuals, its presence in six clinic patients could be explained by a larger number of measurements of ANC, selection bias, possible effects of less frequent alleles or homozygosity (codominant inheritance), alone or in combination.

The results of this study imply that Arabs have two healthy subpopulations with different ANCs. How big these subgroups are is unclear because of a high variability of neutrophil count and a need for multiple sampling in order to properly classify all the subjects. Nonetheless, in clinical practice this requires two cutoff values for neutropenia, one of 1.0 × 10⁹ cells/L and perhaps less for neutropenic population and another of 1.5 × 10⁹ cells/L for non-neutropenic subpopulation. The use of a single value of ANC for defining neutropenia in such a population would also endanger a healthy subpopulation with a higher neutrophil count; it would be put at risk of over-treatment with chemotherapy, under-investigation of neutropenia or under-utilization of antibiotics for infections. Therefore, identification of healthy subjects with lower neutrophil count through screening may be required in this population.

Earlier report suggested that BN in an Arab tribe is inherited as autosomal dominant trait. In that report, however, ANC < 2.0 × 10⁹/L was used as a definition of neutropenia 10. Our analysis of four families using ANC < 1.5 × 10⁹/L as a cutoff value for neutropenia supports autosomal dominant mode as a likely pattern of Mendelian inheritance (Fig 6). However, observation of a profound neutropenia in several subjects (Figs 4 and 5) suggests possibility of the existence of a more than one mutation for BN in this population.

An indication that BN may be inherited as an "ancient" and widely spread trait comes from the presence of neutropenia in diverse Arab tribes (Fig 2). As endogamy was strictly enforced and inter-tribe marriages were infrequent in the past, the presence of neutropenia in many tribes suggests that its presumed gene was present long before tribal partitioning. In the more recent past, a more intense mixing of people than commonly presumed is evidenced by both historical analysis of migrations and high diversity of alpha-thalassemia mutations in these populations 1319. The observed variations in the prevalence of neutropenia, however, could be due to a sample size, founder effect, and different rates of expansion and mixing of populations. The effect of widespread consanguinity (increased homozygosity due to inbreeding) on BN could be determined only after a better phenotype-genotype correlation is established. Finally, that BN may be ubiquitous in Arabs is suggested by a high prevalence of neutropenia in people of African origin; being neighbors they could have exchanged genes during their long and common history. However, the purpose, adaptive value and evolutionary origins of BN still need to be addressed.

It may be possible that the inter-ethnic differences in neutrophil counts may be the consequence of human adaptation to malaria, similar to the thalassemias and hemoglobin S. We base this speculation on three epidemiological observations. First, there is a superimposition of the distribution of Plasmodium falciparum and the geographic distribution of populations with low neutrophil counts. Second, there is an inverse correlation between the intensity of exposure to Plasmodium falciparum and the mean neutrophil count in the same population. Historically (before Columbus), Mexicans were never exposed to malaria and have the highest mean neutrophil count (Fig 7) and the lowest prevalence of neutropenia (0.4%) 2. In contrast, the Arabs and Africans have a long history of endemic malaria and both have low mean neutrophil count and a high prevalence of neutropenia of 10.7% and 4.4%, respectively 12. Similarly, ANCs are lower in Yemeni and Ethiopian Jews, who both reside in regions with malaria and have a high frequency of neutropenia 910. Historically, Europeans had a relatively mild exposure to P. falciparum and their mean neutrophil count and the prevalence of neutropenia (0.8%) are also low. A third observation in support of malaria hypothesis is a direct correlation between the age-specific mortality from malaria and the age-specific prevalence of BN. Mortality from malaria is highest in the first five years of life when the prevalence of neutropenia appears to be highest 2. The mechanism by which a low neutrophil count could provide a survival advantage in malarial infections is unclear. Neutrophils are charged with multiple toxic molecules which could direct destructive processes in the brain and lungs which are the most common causes of deaths due P. falciparum 20. Thus, a lower neutrophil and monocyte count found in subjects with BN may be associated with fewer deaths from malaria by causing a less destructive tissue reaction.