Significant pituitary siderosis is common in transfusion-dependent sickle cell disease Free

TO THE EDITOR:

Iron overload is common in hemoglobinopathies, such as in chronically transfused thalassemia major (TM) and sickle cell disease (SCD). Exposure of major organs to iron results in cirrhosis, heart failure, diabetes, and hypogonadism, which is more prevalent in TM than SCD. Pituitary siderosis occurs in the first decade of life and leads to a 50% prevalence of hypogonadotropic hypogonadism (HH) in patients with TM.¹ Pituitary iron deposition precedes gland volume loss by a decade in TM,² suggesting a window where the clinical effects can be reversed by aggressive iron chelation.³ Patients with SCD have effective erythropoiesis; hence, their exposure to non–transferrin-bound iron (NTBI) and extrahepatic iron burden is less intense than in those with TM.⁴ Nonetheless, pancreatic iron loading is common in transfused patients with SCD by their early 30s,⁵ and cardiac siderosis is observed in heavily iron-loaded subjects.⁶ Pituitary iron and pituitary function have not been studied in SCD. Our goal was to determine the prevalence, predictors, and significance of pituitary siderosis in transfused patients with SCD.

A cross-sectional study was done on patients with SCD who had undergone either clinically or research-indicated brain magnetic resonance imaging (MRI; CCI-14-00034) to compare iron values between 1.5 Tesla (T) and 3.0 T between 2009 and 2022. Informed consent was obtained for the research MRIs, and waiver of informed consent was granted by the Children’s Hospital of Los Angeles institutional review board for the clinically obtained MRIs. MRIs with pituitary volume, R2 (1/T2) in coronal and sagittal planes of the pituitary, liver iron concentration (LIC) by R2∗ (1/T2∗), heart T2∗ (magnetic resonance imaging relaxation parameter), and pancreas R2∗ were included; patients or examinations with missing data were excluded. We obtained patient age, sex, diagnosis, and duration of chronic transfusions through medical record review. Ferritin, transferrin saturation, luteinizing hormone, follicle-stimulating hormone, and testosterone levels were obtained from medical record review for the dates closest to the date of the MRI within 6 months. Pretransfusion hemoglobin S and reticulocyte percentages were averaged over the 6 transfusions preceding the MRI as a marker of marrow suppression by transfusions. We eliminated patients who received bone marrow transplants, because conditioning regimens are known to cause HH.⁷ 

All pituitary MRIs were collected on a 1.5-T Philips Achieva using an 8-element head coil. The anterior pituitary was manually segmented on the first echo time image, and the signal decay curve was fit, voxel wise to a simple monoexponential function without offset. Anterior and posterior pituitary boundaries were manually segmented and summed by method of disks to calculate pituitary volumes. Both pituitary R2 and volumes were referenced to normal values to calculate z scores; sagittal and coronal R2 z scores were averaged together to create a single pituitary “iron” z score.

Univariate and multivariate linear regression was used to determine the relationship between pituitary iron and years of transfusion, ferritin levels, and other organs with iron overload. We dichotomized the pituitary R2 value into high (R2 z > 5) or low (R2 z < 5) risk for hypogonadism based on prior data in thalassemia and rare anemias,¹ demonstrating that a R2 z score >5 conveyed 50% risk for hypogonadism. Using this dichotomization, logistic regression and receiver operating characteristic curves were calculated between pituitary risk and clinical, laboratory and imaging predictors, allowing computation of diagnostic yield (area under the receiver operating characteristic curve [AUROC]) and optimal discriminatory thresholds.

Patient demographics are in Table 1. A total of 41 patients were screened, of which 31 met the inclusion criteria. Patients were aged 18 ± 7.2 years and well distributed by sex. All patients received blood transfusion every 2 to 4 weeks for an average of 9.8 ± 4.6 years. All patients were on oral iron chelation therapy (deferiprone or deferasirox) beginning 6 to 12 months after transfusion initiation, with variable compliance. None of the patients carried a clinical diagnosis of hypogonadism.

On average, patients had severely increased total body iron stores with mean serum ferritin of 5037 ng/mL and LIC of 22.6 mg/g. Transferrin saturation was elevated (62.6% ± 24.2%), indicating increased risk for NTBI.⁸ Five patients (16%) had mean pancreas R2∗ >100 Hz, the threshold for cardiac iron loading.⁹ Of these 5 patients 3 also had MRI-detectable cardiac iron (T2∗ < 20 ms). Remarkably, pituitary iron was detectable in 24 patients (73%), and severe in 9 patients (27%), occurring in the absence of pancreatic or cardiac iron loading in most patients. Although patients with SCD have lower transferrin saturations and NTBI than typically observed in TM, pituitary iron loading was nearly ubiquitous in this high-risk cohort, and reached concerning levels in a large percentage.⁴ 

Although no patient had documented clinical HH, severe pituitary iron deposition was associated with preclinical disease. Figure 1A demonstrates that follicle-stimulating hormone is negatively correlated to R2 z score, with patients having z scores >5 clustered at the lower limits of normal (r = −0.63, P = .04). Figure 1B depicts a similar relationship of pituitary R2 with respect to volume z score (r = −0.62, P = .03). The volume z score in subjects having severe pituitary iron was −1.01 ± 0.94, compared with −0.15 ± 1.05 (P = .04). These data are among the first to clearly document iron-mediated pituitary toxicity in patients with SCD.

Pituitary R2 z score correlated with other markers of iron overload, such as years of transfusion, ferritin, LIC, R2∗ pancreas, and 1/T2∗ heart, with r² values ranging from 0.36 to 0.62 (Table 2), but not hemoglobin S or reticulocyte percentages. Ferritin, pancreas R2∗, and years of transfusion were retained in the multivariate linear regression model with a combined r² value of 0.84. LIC and ferritin were the best univariate predictors of severe pituitary iron loading by logistic regression, with an AUROC of 0.92 and 0.91, respectively. Although LIC is the single best predictor of total body iron, it was correlated with both serum ferritin and years of transfusion. Therefore, the best multivariate predictor was ferritin and years of transfusion therapy, with a combined AUROC of 0.96.

The working model for endocrine and cardiac iron loading suggests that NTBI species circulate whenever chelator is not present in heavily iron-loaded individuals. These NTBI species enter organs constitutively through divalent metal transporters, such as calcium and zinc transporters, unlike the highly regulated transferrin-bound iron uptake.¹⁰ The relative vulnerability of different organs (pituitary > pancreas > heart) likely arises from organ-specific expression of these divalent metal transporters. Molecular characterization of extrahepatic iron transport mechanisms could potentially lead to therapeutic targets; however, compliance to iron chelation and 24/7 control of NTBI species remain the best current treatment options.^3,11 

The study has limitations. The poor control of total body iron stores reflected a clinical selection bias toward patients with SCD at highest risk. The endocrine phenotype was incompletely characterized and would have benefitted from provocative testing. Although it would be interesting to have direct measures of NTBI, elevated transferrin saturation is strongly correlated with NTBI.⁸ 

This work demonstrates 2 key novel findings: (1) Pituitary iron loading is common in heavily transfused patients with SCD and often occurs in isolation from other extrahepatic organ involvement, and (2) Pituitary iron accumulation is associated with secretory dysfunction and smaller gland volumes. Although regular pituitary screening may not be necessary in all transfused patients with SCD, we encourage clinicians to keep a high index of suspicion for iron overload-mediated HH in their adult patients with severe iron overload.