Graves’ Disease and Microcytic Anemia: A Forgotten Connection

Introduction

In the 1970s and 1980s, clinical medicine was more phenomenological than today. Large case series and clinical correlations were the basis for many publications of the period. In large case series of hyperthyroid patients, it was noted that there was a strong correlation with decreased red cell size (microcytosis) [1–3], but no mechanism was proposed. Furthermore, it was known that microcytosis was often associated with iron deficiency [4], although this was not understood mechanistically either. In the case of the microcytosis of hyperthyroidism, there seemed to be no correlation with iron deficiency.

The mechanism of microcytosis in iron deficiency is now well understood and characterized [5]. Iron deficiency in red cell progenitors is associated with heme deficiency, activating the HRI (kinase), which impedes globin translation, and there are global effects on protein translation. The red cells with less hemoglobin are made smaller due to limiting cell components. Hemoglobin consists of tetramers of alpha and beta globin, with each subunit coordinating a separate heme moiety. If iron is deficient, heme becomes deficient, potentially leaving excess globin without its cofactor. This can present a proteostasis problem for the cell due to the excess cofactor-less globin. A regulatory mechanism has evolved to counteract this problem: heme deficiency and/or proteotoxic stress that occurs in the setting of iron deficiency activates HRI, the heme regulated inhibitor. This leads to phosphorylation of eIF2alpha, which inhibits formation of the ternary complex (eIF2-GTP-Met-tRNA), blocking translation of abundant mRNAs, including globin. This regulatory mechanism thus normalizes cellular heme and globin levels, but at the expense of making smaller red cells [5].

An alternative mechanism of microcytosis involves overproduction of the iron regulatory signaling compound hepcidin. Hepcidin is a mediator of global iron metabolism that responds to input from iron availability and from cytokines such as IL6. In the presence of inflammation, hepcidin production may increase, leading to degradation of ferroportin; plasma iron then falls, leading to an apparent iron deficiency state. In the setting of hyperthyroidism, however, plasma iron and iron saturation are generally unchanged or increased, making a role for hepcidin in the microcytosis of hyperthyroidism unlikely.

Case Report

Our patient was found to have a microcytic anemia that was poorly explained by the “usual suspects”: iron deficiency or inflammation. He was a 56-year-old man who presented to the Hematology clinic at the Corporal Michael J. Crescenz VAMC with an illness beginning in September 2020 that became progressively worse, characterized by weight loss, lightheadedness, palpitations, anxiety, and malaise. He also described a sensation of his eyes being difficult to open. He had a chronic outward strabismus of the left eye, which may not be relevant to the current illness. He had no evidence of blood loss – no history of blood loss, no iron deficiency, and no prior transfusions of red blood cells. The patient likewise had no evidence of an iron absorptive defect – no history of gastric or abdominal surgery, and no GI symptoms. Importantly, there was no evidence of iron deficiency per se, as ferritin levels were high before methimazole treatment.

Physical exam showed temperature 37°C, pulse 100, respiration 20, blood pressure 151/61. He was thin appearing with normal skin and normal mucosae. The thyroid was diffusely enlarged, although there was no bruit and no palpable nodularity. He had no proptosis or eyelid lag, and extraocular motions were normal, except for the strabismus of the left eye. There was no adenopathy in the cervical, supraclavicular, or axillary regions. The chest was clear, the heart was without murmurs or gallops, and the abdominal exam was benign, without tenderness or enlargement of the liver or spleen. The skin was dry without bruising, and extremities showed no edema. The patient was seen originally as a consultation in the Hematology clinic for microcytic anemia, and diagnostic tests were ordered to elucidate possible causes: hemoglobin, hematocrit, mean corpuscular volume, serum iron, serum iron saturation, and serum ferritin. Thyroid function tests were ordered only after an astute endocrinologist considered that the patient’s symptom constellation could be due to thyroid disease. Thyroid function tests came back markedly elevated, showing biochemical hyperthyroidism (TSH less than 0.015, FT4 of 4.10, and TT3 of 4.75). In addition, TRAb (thyroid receptor antibody) was elevated at 8.18 IU/L (normal range: 0–1.75) and TSI (thyroid stimulating immunoglobulin) was elevated at 8.51 IU/L (normal range: 0–0.55), consistent with a diagnosis of Graves’ disease.

Various imaging tests were performed – a bedside ultrasound of the thyroid showed a markedly enlarged thyroid with increased vascularity and no nodules. Radioactive iodine uptake was markedly increased at 56.4% (normal 4–15%) and 75.7% (normal 10–32%) at 4 and 24 h, respectively. The increased tracer in the thyroid appeared to be uniformly distributed. A mediastinal mass of about 3×3.2 cm was noted, but there was no increase in uptake within the mass. CT of the mass exhibited a triangular configuration and demonstrated diffuse signal dropout on out-of-phase imaging, due to the presence of microscopic lipid. The presumed diagnosis for the mass was thymic hyperplasia. Interestingly, following treatment of the hyperthyroidism, the mass showed near complete resolution on a repeat CT 8 months later.

The clinical course was as follows. The patient was started on methimazole at 20 mg/d. Methimazole exerts its therapeutic effect by inhibiting thyroperoxidase, a crucial enzyme in synthesizing thyroid hormones. This effect decreases the synthesis of thyroid hormones, thyroxine (T4) and triiodothyronine (T3), restoring normal thyroid function. He felt better within a week, and after several weeks, symptoms were completely resolved. His tremulousness was gone, he gained weight, and he recovered a sense of well-being. Thyroid parameters were also improved, and after 4 months of treatment, TSH, FT4, and TT3 had normalized (3.43, 0.68, 0.93, respectively). Methimazole was titrated down to 7.5 mg/d. The pathological antibodies characteristic of the Graves’ diagnosis resolved as well, with TRab returning to less than 1.10 IU/L and TSI returning to 0.45 IU/L, both within normal limits.

Regarding the association of hematology with thyroid disease, we were particularly interested in the changes in hematologic parameters that occurred during the clinical course. A graph of the blood count values shows that mild anemia (Figure 1A) and microcytosis or small red cells (Figure 1B) developed concurrently with the patient’s thyroid symptomatology. As such, a mild degree of anemia would by itself be unlikely to induce any symptoms; we concluded that the thyroid disease caused the symptoms, and the microcytic anemia was a derivative sign. After the diagnosis of hyperthyroidism was made based on biochemical values of excess T3 and TSH suppression, he was started on methimazole 20 mg/d in March 2021. Remarkably, within a few weeks, the symptoms reversed, concurrent with improvement of total T3 (Figure 2). Interestingly, the curves of the hemoglobin (Hgb) versus time reached a minimum at 9.8 g/dL and started upward, as did the curve of the mean corpuscular volume (MCV), which reached a nadir of 76 femtoliters and then turned upward. Over the next 2 months, while the thyrotoxicosis was under treatment, Hgb and MCV returned completely to normal (Figure 1). Thus, the onset and resolution of the symptoms related to T3 excess was correlated very closely with changes in the blood, suggesting a causal effect. Other clinical data showed that the red cell distribution width (RDW) remained low at 13–14 and then rapidly increased to 17 at the time of initiation of the hyperthyroid treatment with methimazole. These findings suggested that thyrotoxicosis was leading to the production of small erythroid precursor cells, and when this was reversed, the mixture of small cells together with corrected larger cells resulted in a mixed population with high RDW (Figure 3).

The connection to iron homeostasis is also interesting. An initial hypothesis was that the microcytic anemia could be due to iron deficiency. However, serum ferritin levels in general correlate quite well with total body iron, and the ferritin levels of our patient were initially very high at the time of diagnosis (821–844 ng/mL, normal range 24–336 ng/ml), and they remained high until treatment was initiated. Only after treatment was initiated did the ferritin levels return to normal (287–221 ng/mL) (Figure 4A). These data are consistent with T3 at supraphysiological concentrations initiating a block in iron utilization and causing iron to accumulate as ferritin, leading to high serum ferritin levels. Iron stores as judged by serum ferritin appeared to be adequate, or more than adequate throughout the time course, indicating that iron deficiency per se was not the cause of microcytic anemia. Indeed, iron levels (purple diamonds, mcg/dL) and iron saturation (red diamonds, %) jumped up following initiation of methimazole. These data are consistent with the notion that hyperthyroidism initiated a functional block in iron utilization that was alleviated following treatment (Figure 4B). Hepcidin excess might account for a block in iron utilization such as occurs in the anemia of inflammation, but this is invariably associated with a fall in serum iron and a fall in iron saturation of transferrin. In the anemia of hyperthyroidism by contrast, serum iron and iron saturation remain normal (Figure 4), making hepcidin excess an unlikely explanation for the microcytosis of hyperthyroidism.

Discussion

The current literature is almost devoid of reports of hyperthyroidism causing microcytic anemia, with only 1 retrospective descriptive report from 2010, showing microcytosis in 87.7% of 235 patients with hyperthyroidism. In these cases, the clinical courses were not reported, so we do not know if the microcytic anemia remitted concurrently with remission of the hyperthyroidism. Moreover, no mechanism was proposed [6]. However, the subject was thoroughly explored and documented 40–60 years ago, but without mechanistic insights. In a paper called, “Red cell changes in hyperthyroidism” [2], the investigators prospectively studied 100 untreated non-anemic hyperthyroid patients. The most significant finding was a low mean corpuscular volume (MCV), which was invariably present throughout the hyperthyroid state. Treatment with beta-adrenergic blocking drugs did not significantly alter any of the red cell parameters. On the other hand, the MCV increased and was restored to normal with radioiodine or carbimazole treatment, which are alternative treatments for hyperthyroidism. There was no significant difference in Hgb, PCV (packed cell volume), and MCHC (mean corpuscular hemoglobin concentration) of the untreated hyperthyroid patients and normal controls. However, the MCV of the hyperthyroid patients (80.5±3.38 fL in females and 82.06±3 fL in males) was significantly lower than controls (87.0±4 fL in females and 88.0±4 fL in males, with a P value of <0.001). A suggestion was made that the MCV may have a useful role in the diagnosis of hyperthyroidism. Review of very old literature reveals that as early as 1927 Jorgensen and Warburg described microcytosis in Graves’ disease [7], although the mechanism for the microcytosis was not understood. An instructive study of 19 thyrotoxic patients that underwent bone marrow biopsies before and after treatment was performed by Lahtinen in 1980 [3]. Here too the MCV was significantly lower before treatment than in the euthyroid state (P<0.001). The saturation of transferrin was relatively high (greater than 60%) in 5 patients before therapy and this normalized with therapy. The striking finding was that the proportion of erythroblasts with iron deposits, termed sideroblasts, was significantly higher in the hyperthyroid than in the euthyroid state. The proportion of sideroblasts was over 50% in 10 out of the 19 patients. Coarse perinuclear iron granules were found in sideroblasts of 7 patients in the hyperthyroid phase but none in the euthyroid phase, suggesting that iron utilization by the red cell precursors was disrupted, leading to accumulation of iron in unused ferritin granules within the cells. It was not clear from the studies whether the deposits were mitochondrial or cytoplasmic, since electron micrography was not performed. However, this resembles functional iron deficiency such as occurs in states of ineffective erythropoiesis (eg, thalassemias). A paper from 1969 by Rivlin and Wagner presented ferrokinetic data on 4 hyperthyroid patients conducted both before and after antithyroid therapy [8]. The mean red cell survival recorded in the 2 groups did not differ. In the hyperthyroid group, turnover of plasma iron was faster, but iron incorporation was less complete, suggesting an impairment of utilization of iron, as occurs in ineffective erythropoiesis. A block or attenuation of globin mRNA translation as occurs in thalassemia might account for this phenotype (see below).

Conclusions

Regarding a proposed mechanism for microcytosis occurring in the setting of hyperthyroidism, a mechanism for the functional iron deficiency leading to accumulation of sideroblastic granules in red cell precursors and microcytosis in the hyperthyroid state was suggested by a more modern study of the biochemistry of hyperthyroidism using HeLa cells [9]. These investigators showed that T3 hormone at supraphysiological concentrations activates integrated stress response pathways (ISR) that induce stress granule formation and ER-mediated stress pathways, as well as activating ISR mediating kinases. They studied effects on PKR (protein kinase R), which usually responds to double-stranded RNA and viruses. However, in red cell precursors, HRI (heme regulated inhibitor) might be the true downstream target of the supraphysiological concentrations of T3, given that HRI is the most highly expressed form of ISR signaling kinase in those cells. HRI is known to mediate phosphorylation of eIF2alpha, thereby blocking recycling of GTP/GDP by eIF2, decreasing the concentration of the ternary translation complex and attenuating translation of abundant mRNAs such as globin mRNA (Figure 5). The failure to translate globin mRNA in red cell precursors would be expected to reduce iron utilization such as occurs in beta thalassemia, explaining the appearance of heavily granulated sideroblasts in hyperthyroid patients.

To complete the mechanistic picture, we needed to establish a link of the supraphysiological concentrations of T3, as it occurs in thyrotoxicosis, to phosphorylation of eIF2alpha in red cell precursors. A recent paper has uncovered novel principles of thyroxine function, showing that T3 at high concentrations activates the ISR signal transduction pathway [9]. This effect may be mediated via stress granule production or disruption of protein proteostasis due to disruption of translational contexts. In turn, these effects may activate 1 or more of the 4 protein kinases that lie upstream of eIF2alpha phosphorylation [10,11]. The 4 kinases generally mediate distinct stress signals (Figure 5). GCN2 responds to excess of uncharged tRNA switching between translation of short upstream ORFs and major mRNA depending on availability of apo-tRNAs. PERK responds to ER stress; the kinase domain resides inside the ER, shielded by chaperones. When the unfolded proteins in this compartment overwhelm the chaperones, then the ER domain signals to the kinase domain in the cytoplasm, leading to phosphorylation of eIF2alpha. PKR is a kinase apparently designed to respond to dsRNAs, as occurs in some viral infections. The result is the same in all cases, which is the phosphorylation of eIF2alpha and down-titration of the ternary complex, consequently attenuating global translation. Supraphysiological concentrations of T3 can act through PKR to attenuate translation. HRI (heme regulated inhibitor) is the fourth relevant kinase. In the erythroid lineage HRI expression increases during terminal erythropoiesis. Starting at the basophilic erythroblast stage, HRI is the predominant eIF2alplha kinase response for 90% of eIF2 alpha kinase activity. It is expressed at levels that are 2 orders of magnitude higher than the other 3 eIF2alpha kinases. Thus, while it is formally possible that the T3 signal to ISR comes through PRK, it is much more likely that HRI is the true mediator. HRI often acts as a heme dependent kinase, responding to situations in which heme is deficient, and thus globin is present in excess. HRI is activated during heme deficiency to phosphorylate eIF2alpha, which simultaneously inhibits translation of globin messenger RNA. Iron deficiency impacts heme levels and stimulates HRI, leading to attenuation of globin translation and better matched production of heme and globin polymers. The situation in hyperthyroidism is somewhat different (Figure 5). Iron deficiency and heme deficiency are not present, although functional iron misutilization is present, as suggested by ferrokinetic data [8]. Instead, HRI may respond to a different stress signal. High levels of T3 may induce stress granule production and proteotoxic stress, which may activate HRI and/or PKR. The output of the activation of the kinases, termed the ISR, is to slow translation of abundant messenger mRNAs such as that of globin in the red cell precursors. Interestingly the control of red cell size is tightly linked to globin mRNA translation, through cell cycle checkpoint effects. Thus, the decreased red cell globin translation leads to smaller red cells (lower MCV), such as occurs in thalassemia [5]. Conversely, when thyroid function is restored to normal with methimazole treatment, the stress induced HRI kinase activity is normalized, eIF2 alpha phosphorylation is reversed; GDP-GTP exchange occurs, and globin mRNA translation being restored red cell size reverts to normal (Figure 5).

The strengths of this case report are that we were able to observe the entire clinical course from presentation to diagnosis, to treatment to resolution, and thus the correlations of T3 with hemoglobin and MCV could be observed over time. Another strength is that we have also been able to propose a mechanism for the hematology effects of supraphysiologic concentrations of thyroid hormone, incorporating modern concepts of environmental signaling through the ISR kinases and eIF2 phosphorylation. A limitation of the study is that direct demonstration of activation of the HRI kinase in the patient’s red cell precursors is lacking but could be addressed in future research. A key although non-trivial experiment could be to measure eIF2 alpha phosphorylation and globin mRNA translation activity in red cell precursors from the bone marrow of a patient with hyperthyroidism/Graves’ disease. The prediction is that phosphorylation would be increased and globin translation would be decreased in patients with hyperthyroidism compared with controls or patients with corrected hyperthyroidism.