The first clues suggesting a potential involvement of microcytic anaemias in psychiatric disorders date back to the 1980s. The Italian woman with bipolar affective disorder and thalassaemia minor was ascertained in Canada in 1986 [1]. Her family history of potential co-segregation of the mood and the blood disorders raised the hypothesis of linkage between the ß-globin gene and a gene for manic-depressive illness on chromosome 11p15. At that time, chromosome 11p15 had become popular in psychiatric genetics after the first linkage study in the Amish population [29, 30]. The subsequently published cases often reported co-segregation in affected relatives, and most of the authors hypothesised that linkage disequilibrium between globin genes and genes involved in neurotransmission was a possible explanation. They underscored, for example, that the ß-globin gene is closely linked to the genes encoding tyrosine hydroxylase or dopamine D4 receptors on chromosome 11p15. However, the quality of family data was limited in several of these early reports due to the lack of haematological information regarding some relatives, the lack of molecular diagnosis of the thalassaemic genes, and the heterogeneity of the psychiatric cases involved.

Moreover, given the high worldwide prevalence of mood disorders, co-occurrence with microcytic anaemias might be attributed to chance, especially in areas where thalassaemia genetic mutations are highly prevalent. Nevertheless, some observations remain very intriguing, such as the multiplex family ascertained in Japan [3], where thalassaemias are very rare.

Our first observation of the potential association between microcytic anaemias and bipolar spectrum disorder [11] was serendipitous. We collected blood counts to calculate glucose-6-phosphate dehydrogenase (G6PD) activity in an early study using G6PD deficiency as a phenotype marker of X-linkage with manic-depressive illness [30]. We then noticed the disproportion of reduced MCV among some diagnostic subgroups, namely psychotic manias [11]. When we discussed our results with the late professor Antonio Cao, a distinguished expert in thalassaemia, he warned us about geographical stratification as a potentially misleading factor. Indeed, a wide variation in the prevalence of ß-thalassaemic carriers (3.8 to 38%) was found across Sardinian areas by the 1966’s pioneer study of more than 6000 observations in 52 villages showing the same distribution once observed for malarial morbidity [31]. If mood disorders with psychotic and/or manic features are more prevalent for unknown reasons in once malarial areas in Sardinia, a spurious association with heterozygous thalassaemias might have derived in patients from our catchment area. A recent survey on 63,285 13- to 14-year-old students has found a more homogeneous prevalence distribution of ß-thalassaemic carriers (9.1 to 11.7%) across 13 Sardinian sub-regions [32]. Based on data of Hb A₂ concentration, we had estimated a 16.4% prevalence of heterozygous β-thalassaemia in bipolar subgroups from our previous survey [13], which is in any case higher than the 10.3% mean prevalence currently found in the general population. We are not able to compare data regarding prevalence of α-thalassaemic carriers because the 80 μ³MCV cut-off we choose to define microcytic anaemias does not detect many carriers, particularly those with three α-globin functional genes (- α / αα). This may also be one of the reasons why there are fewer case reports in the literature and fewer multiplex sibships from our study, despite the much higher prevalence of α- than ß-thalassaemic carriers in the population (10 to 37% in the recent Sardinian survey [32]).

We have limited our study and this discussion to the simplest cases because rarer haemoglobin mutations or complex phenotype interactions can go undetected by common screening methods.

Published surveys of microcytic anaemias and/or heterozygous thalassaemias in psychiatric populations (Table 2) vary in terms of sample size, diagnostic methods, and choice of control subjects. However, we emphasise that they are invariably in the same direction, even if single studies were not invariably able to conclude for a significant association. We have calculated that the excessive prevalence of heterozygous ß-thalassaemia found in the Argentinean [12] and the Iranian [14] bipolar subgroups compared to controls (7.8% versus 1.9%; 8.2% versus 4.2%; respectively) would have attained statistical significance with larger (for example threefold) sample sizes. The authors reported an estimated prevalence of ß-thalassaemic carriers in the general population of 5% in Argentina [12] and 10% in Mazandaran, the northern province of Iran [14]. We maintain that the association has to be researched in areas with high prevalence. Thalassaemia genes are common in many populations of the Mediterranean, Sub-Saharan Africa, Middle East, India, China, and South-East Asian regions that have previously shown endemicity of malaria. On the contrary, surveys carried out in other countries will be less likely to reveal significant results. For example, in a population of 289 patients who were consecutively admitted to a public psychiatric hospital in California and screened for physical disorder, 13 cases (4.5%) were found with unspecified anaemia, but only one (1.1%) with thalassaemia [33]. Nevertheless, people with globin genetic mutations may end up anywhere given the increasingly mobile world, as witnessed by the variety of countries where cases of concurrent heterozygous thalassaemias and mood disorder were ascertained (Tables 1 and 2).

In our present replication, we confirmed that microcytic anaemia could be found in up to ¼ of Sardinian outpatients with recurrent mood disorders. Once more, manic schizoaffective disorder is the RDC diagnostic subgroup with the highest prevalence, but the new category of hyperthymic depression has emerged as potentially associated with the microcytic anaemia. The bipolar/non-bipolar distinction is not as clear-cut as previously assumed in our previous studies [11, 13]. For example, in the present survey, “typical” bipolar cases (RDC diagnosis of bipolar with mania) had a lower prevalence of microcytic anaemias (15.6%) compared to the remaining subgroups, including non-bipolar categories (Table 4). One possibility is that patients referred to our clinics over the last 15 years differ from those included in our previous surveys in terms of the severity of depressive episodes. Indeed, microcytic anaemias may have a role also in non-bipolar cases. For example, one of the early reports about the small population of Mediterranean women hospitalised in Germany [10] found that heterozygous ß-thalassaemia was associated with peculiar symptoms (muscular weakness and bone pain) that may be prominent in severe depressive syndromes. Moreover, one of the case reports (Table 1) regarded a Belgian woman with severe recurrent depression [7].

Our is a clinical pharmacology unit that might select patients with some characteristics that make them different from common psychiatric outpatient units. Lithium may be prescribed by any psychiatrist in the Cagliari area, but our unit has been one of the reference centres for the monitoring of lithium treatment since its introduction in Sardinia. There might have been a selection of patients responding to lithium during acute episodes or carrying high suicide risk. Our interest in genetic studies, dating back to the 1980s, has led us to adopt modified RDC criteria rather than DSM or ICD criteria. With regard to the thalassaemic trait, it is mostly asymptomatic, and patients are not aware of their presence unless they had undergone some screening procedure. We often detect it by simply examining routine blood counts. In any case, we do not choose a specific psychiatric treatment. However, we have learned that β-thalassaemia heterozygotes with Manic Schizoaffective Disorder appear to respond well to lithium alone or combined with low dose antipsychotics, whereas β-thalassaemia heterozygotes with predominant depressive episodes (Bipolar with Hypomania, Recurrent Major Depression, or Hyperthymic Depression) are often stabilised with lithium alone or combined with antidepressants. However, whenever possible, antidepressants are discontinued to prevent switch to hypomania or cycle acceleration. The latter caution is particularly recommended to prevent mood instability in cases with Hyperthymic Depression, the so-called “bipolar IV” according to Akiskal and Pinto [23].

Surveys of psychiatric morbidity in populations of β-thalassaemia heterozygotes (Table 3) show that there may be a particular susceptibility to depressive symptoms. We choose to include in this review some studies on relatives of children with Cooley’s anaemia who had undergone a psychiatric assessment because they represent samples of ß-thalassaemia carriers even if data were collected for another aim. The advantage is that no further molecular diagnosis is necessary. However, such studies were focused on depressive symptoms that are the most plausible to be found in families of severely ill children or at risk of giving birth to ill children. We can speculate that bipolarity did not emerge as often happens in non-clinical samples. Carta and Angst [34] already warned that epidemiological studies, especially when derived from interviews carried out by lay staff, which only reflect the patient's point of view, may miss the detection of hypomanic episodes. More recent surveys summarized in Table 3 started from subjects diagnosed with heterozygous ß-thalassaemia using common screening methods (mostly haemoglobin electrophoresis) [18, 20, 21].

The negative survey from Iran [18] was limited to depression and a relatively young age cohort. It found a small higher prevalence in the 208 β-thalassaemia heterozygotes (47.1%) compared to the 200 controls (43.5%). Given those rates, the statistical difference would have been achieved only with a much larger (e.g., ten-fold) sample.

On the contrary, the most striking evidence of the role of heterozygous thalassaemia as a predisposing factor to severe mood disorder is provided by the recent Sicilian study [21]. The aim was to follow-up β-thalassaemia heterozygotes (compared to a control group) by studying mortality and hospitalisations for various complications. The following causes of hospitalisation were investigated: arrhythmia, cholelithiasis, cirrhosis, diabetes, mood disorders, kidney diseases, and ischaemic cardiomyopathy. Among other results, there was an increased risk of hospitalisation for mood disorders of 2.08 (95% CI 1.15 to 3.75; P = 0.015) in β-thalassaemia heterozygotes. The Sicilian study has many advantages. The sample size was the largest studied to date, including almost 5000 subjects with heterozygous ß-thalassaemia. Another advantage in making bias less probable is that the authors do not appear to be aware of all previous studies associating heterozygous thalassaemias with psychiatric disorders. Indeed, psychiatric illness was just one of the potential health problems investigated, and psychiatric assessment was independent of the authors’ judgement because it was based on data of subsequent hospitalisations. Finally, the two-fold odds ratio was just mentioned without any particular emphasis or any hypothesis on potential pathogenic mechanisms. One limitation is that no specific distinction was provided about hospitalisations for manic or depressive episodes.

Co-segregation in relatives is considered a robust method to rule out false associations due to geographical stratification. To date, the evidence regarding the role of thalassaemic genes in mood disorders cannot be considered as established. Early reports were based on limited numbers of probands and affected relatives. The latter were not studied directly in many cases, or their haematological phenotype was unknown. Nevertheless, some observations remain very intriguing, such as the multiplex family ascertained in Japan [3], where thalassaemias are very rare.

Our study of 30 probands and their 45 affected siblings adds some information. The 78% rate of phenotype concordance is striking even if not complete. Nevertheless, we must consider that 50% of siblings of thalassaemic carriers are expected in advance to be haematologically identical by descent with the proband, irrespective of their psychiatric status.

Apart from the three sibships segregating for the α-thalassemia deletion, we did not deepen the diagnosis at the molecular level. Molecular diagnosis is mandatory in the prevention of Cooley’s anaemia because genetic compounds are expected to be frequent in Sardinia. For example, coinheritance of mutated ß and α globin can result in normal MCV, because ß and α chains are less unbalanced compared to simple heterozygotes. Therefore, we cannot exclude that some phenotypically-discordant affected siblings may represent false negatives. On the other hand, phenotypically-concordant siblings may carry different genes compared to the proband.

The predominant mutation of the ß-thalassaemia chromosomes in the Sardinian population is the codon 39 nonsense mutation, which accounts for 95.7% of cases [35]. On the contrary, the α-globin genotypes identified with thalassaemia screening in Sardinia are known to be complex [36, 37].

We would also like to underscore that some of the 10 discordant siblings carry additional mutations potentially involved in the pathogenesis of psychiatric disorders, such as G6PD deficiency [38, 39].

With regard to the concept of the bipolar spectrum, we considered 20 siblings with unipolar depression as affected. The bipolar/non-bipolar distinction has long been challenged by family studies of bipolar probands. Increased rates of depression are found in relatives of both bipolar and unipolar probands compared to normal probands when incapacitating episodes alone are considered [40]. This is the reason why we included incapacitation in the major life role during depressive episodes among criteria for affected siblings. It is noteworthy that one of the reported cases in the literature regarded a Belgian woman with severe recurrent depression and co-occurrence of heterozygous β-thalassaemia and recurrent major depression in the proband’s father and two paternal uncles.

The heterogeneity of phenotypic presentation is not unusual in psychiatric genetics: in fact, examples exist of identified mutations that predispose to a variety of neurological or neuropsychiatric phenotypes [41]. Therefore, reports of the potential role of heterozygous thalassaemias in various psychiatric illnesses as those summarized in Table 1 are not unexpected, including the Angolan woman and the Belgian man hospitalised in Switzerland for exacerbation of paranoid schizophrenia with co-segregation of heterozygous β-thalassaemia and schizophrenia in several relatives [9].

Since the first suggestion of a potential role of heterozygous thalassaemias in psychiatric disorders, various potential mechanisms have been hypothesised. Altinoz and Ince [42] have recently reviewed the emerging roles of haemoglobins in mental disorders, discussing thoroughly metabolical, genetical, and immunological aspects. We also suggest that increased concentrations of unconjugated hyperbilirubinemia are included, as these may be a shared characteristic of heterozygous thalassaemias, G6PD deficiency, and Gilbert’s syndrome.