Induction of Heme Oxygenase-1 and Ferritin in the Kidney in Warm Antibody Hemolytic Anemia

Abstract

Warm antibody autoimmune hemolytic anemia is usually associated with extravascular hemolysis. We report a case of a 42 year old man with sustained and moderately severe warm antibody autoimmune hemolytic anemia, hemoglobinuria, hemosiderinuria, and acute kidney injury (AKI). We demonstrate marked induction of heme oxygenase-1 (HO-1) and increased expression of ferritin in renal tubules along with increased deposition of iron in renal proximal tubules. These findings in this clinical case thus recapitulate those observed in experimental models of heme protein-induced kidney injury in which a coupled induction of HO-1 and ferritin occurs in the kidney. We discuss the pathobiologic significance of these findings and suggest that this linked response confers cytoprotection to the kidney exposed to hemoglobin and mitigates the severity of acute kidney injury that may otherwise occur. Finally, this case report documents that nephrotic range proteinuria can occur in autoimmune hemolytic anemia complicated by hemoglobinuria.

INTRODUCTION

Acute kidney injury (AKI) is an uncommon complication of hemolytic anemia, and is more likely to occur if the kidney is exposed to large amounts of hemoglobin, or if there are concomitant factors such as volume depletion, acidosis, other nephrotoxins, or sepsis.^1–3 Hemoglobin and other heme proteins such as myoglobin can contribute to AKI through the following pathogenetic processes: (i) renal vasoconstriction, (ii) direct tubular toxicity, (iii) urinary cast formation.^1–3

Studies in disease models demonstrate that the kidney and other tissues can adapt to the exposure of large amounts of heme proteins by inducing heme oxygenase-1 (HO-1) and ferritin.^4–6 HO is the enzyme overwhelmingly responsible for metabolizing heme, converting it to biliverdin, and in the process, iron is released and carbon monoxide is produced.^7,8 The extent to which renal induction of HO-1 and ferritin occurs in human disease in which the kidney is exposed to large amounts of heme proteins merits documentation and investigation.

We report a case of sustained and moderately severe autoimmune hemolytic anemia attended by AKI, and in which we demonstrate induction of HO-1 and ferritin in the kidney.

CASE REPORT

A 42 year-old white man had a 9 months history of progressive weakness, loss of energy, and dyspnea. Evaluation by his primary care physician in February 2004 revealed a Coomb’s-positive hemolytic anemia. His past medical history was unremarkable and, at the time, there were no prescribed medications, in particular, there was no history of use of nonsteroidal anti-inflammatory agents. No exacerbation with cold exposure was seen. Laboratory evaluation showed a hemoglobin of 5.6 g/dL, a warm-antibody mediated hemolytic anemia, and a hypercellular marrow on bone marrow biopsy. Serum creatinine was 1.2 mg/dl (106 μmol/L), and estimated glomerular filtration rate (GFR) 71ml/min/m² (1.18 mL/s) using the MDRD study equation. Urinalysis showed 4–6 red blood cells per high power field, with no red cell casts. Urine protein concentration was 100 mg/dl (1g/L). Serum protein electrophoresis did not show a monoclonal peak and hemoglobin electrophoresis was normal. Antinuclear antibody, Mycoplasma IgM and rheumatoid factor were negative. At diagnosis, he was started on oral prednisone 120 mg a day. Approximately two and a half months after the diagnosis he underwent a splenectomy with removal of a 396 g spleen which on microscopic examination was reported as consistent with ongoing hemolytic anemia. At the time of splenectomy, prednisone was discontinued and he was started on Danazol, 200 mg twice a day, orally, which he took for approximately 6 weeks. Danazol was then stopped because of worsening liver function tests. During this period of time there was no improvement in his overall status with the patient requiring red blood cell transfusions on four occasions. In August 2004 the patient reported development of painless macroscopic hematuria, although he also recorded having seen dark urine since the beginning of his symptoms in February 2004. At the end of September 2004, the patient started treatment with rituximab, 375 mg/m² weekly for 4 weeks, with the last dose of rituximab given just over a week prior to presentation at Mayo on November 9, 2004. At the time of his visit at Mayo, the patient was taking prednisone 15 mg a day, orally, which he had started on his own approximately 6 weeks previously. On physical exam, the patient was afebrile and appeared ill. His blood pressure was 130/70 mmHg without orthostasis, and the pulse rate regular at 90 beats/min. There was a diffuse yellow discoloration of the skin along with scleral icterus, an absence of adenopathy, and the presence of trace pedal edema.

Laboratory evaluation showed a hemoglobin of 7.2 g/dL, a reticulocyte count of 18%, increased LDH at 3250 U/L (122–222 U/L), and the haptoglobin was less than 14 mg/dl (30–200 mg/dl; SI 0.2–2.0 g/L). The peripheral smear showed moderate anisopoikilocytosis with moderate to marked spherocytosis, polychromasia, and circulating nucleated red cells; platelets appeared normal (Figure 1). No red cell inclusions or schistocytes were seen. The Coombs test was positive for IgG and negative for complement. Flow cytometry for surface expression of CD14 and CD59 was negative for paroxysmal nocturnal hemoglobinuria. The serum creatinine and blood urea nitrogen were 1.7 mg/dL (150 μmol/L) and 20 mg/dL (7.1 mmol/L) respectively. Urinalysis showed 4 – 10 RBCs (< 25% dysmorphic), white blood cells, renal epithelial cells, granular and fatty casts, and oval fat bodies; hemosiderinuria and large amounts of urinary hemoglobin were present. The creatinine clearance was determined at 71 mL/min/m² (1.18 mL/s) while 24-hour urine collection showed proteinuria of 3.5g/24h (35g/L). A renal biopsy demonstrated tubular epithelial cell attenuation consistent with acute tubular necrosis; renal tubules also demonstrated prominent amounts of intracytoplasmic iron, and occasional protein reabsorption droplets. The interstitium showed no inflammation, and focal and minimal amounts of fibrosis. The glomeruli appeared normal apart from segmental ischemic wrinkling of capillaries. There was no evidence of immune complex deposition. Electron microscopy demonstrated segmental areas of effacement of podocyte foot process especially over areas of ischemic wrinkling of the capillaries.

Figure 1.

Peripheral smear showed anisopoikilocytosis, spherocytosis, and polychromasia (arrows, 600X).

Formalin fixed, paraffin-embedded tissue sections were stained with Gomori’s Prussian blue to detect ferric iron. Deparaffinized and hydrated tissue sections were placed in a potassium ferrocyanide (10%) and hydrochloric acid (20%) solution, rinsed and counterstained with carbol fuchsin.⁹ For immunohistochemical localization, formalin fixed, paraffin-embedded tissue sections were stained for HO-1 using a polyclonal antibody (SPA-895, Stressgen) as the primary antibody. For ferritin, the primary antibody consisted of anti-rat ferritin obtained from Dr. Hamish Munro, Massachusetts Institute of Technology, Cambridge, MA.¹⁰ A horseradish peroxidase conjugated secondary antibody (SAB-300, Stressgen, Ann Arbor, MI) and diaminobenzidine as substrate for localization were used as previously described.¹¹ In renal tubules, immunohistochemistry demonstrated prominent expression of HO-1 and ferritin (Figures 2A-D) while Prussian blue staining revealed extensive iron deposition (Figure 3).

Figure 2.

Immunohistochemical analysis of Heme Oxygenase-1 (HO-1) and Ferritin in the Kidney. Panel A: Control: Section of the kidney without immunohistochemical analysis (unstained). Panel B: Control: Immunohistochemical analysis of kidney section incubated with normal rabbit serum (in place of primary antibody). Panel C: Immunohistochemical analysis of kidney section incubated with rabbit polyclonal HO-1 antibody (primary antibody). Panel D: Immunohistochemical analysis of kidney section incubated with rabbit polyclonal ferritin antibody (primary antibody). As shown in the negative control sections (Panels A and B), some tubules exhibit punctate brown staining due to hemosiderin deposits but an absence of diffuse brown cellular staining. In contrast, renal tubules exhibit, in addition to this punctate brown staining, diffuse brown staining in the presence of the HO-1 antibody (Panel C), or the ferritin antibody (Panel D). Panels C and D also demonstrate that the endothelium of the peritubular and glomerular capillaries do not exhibit brown staining, thus reflecting a specificity of such staining for renal tubules. Original magnification 100X.

Figure 3.

Prussian blue staining demonstrating large amounts of iron stored in the kidney as hemosiderin (arrows, 600X).

A diagnosis of heme-pigment-induced AKI was made, and because of the proteinuria, therapy with an angiotensin receptor blocker (ARB) was initiated. At the beginning of December 2004, Cytoxan (100 mg PO daily) was prescribed in an attempt to suppress immune-mediated hemolysis. Follow-up of kidney function demonstrated that the serum creatinine had decreased to 1.3, 1.2, and 1.0 mg/dl (115 μmol/L, 106 μmol/L, and 88 μmol/L) at approximately 2 weeks, 1 month, and one year respectively after the initial evaluation for AKI. Following the introduction of Cytoxan, the hemoglobin gradually increased and was 14.5 g/dl one year after initiation of Cytoxan. In January 2006, Cytoxan was reduced to 50 mg, orally, once a day, down to 25 mg once a day in February, and discontinued altogether in March 2006. Since initiation on Cytoxan the patient has not had any more episodes of cola-colored urine providing significant supportive evidence for an immune-mediated mechanism of hemolysis as the predominant cause of red cell destruction in this case. His most recent hemoglobin was 16.2 g/dl with urinary albumin/creatinine ratio of 2 mg/g (<17 mg/g; March 2008). Table 1.

Table 1.

Clinical course

06/2003 – Starts developing weakness, loss of energy, and dyspnea.

02/2004 – Diagnosis of Coomb’s-positive hemolytic anemia; starts on oral prednisone 120 mg a day.

05/2004 – Undergoes splenectomy; prednisone is discontinued; starts Danazol 200 mg po b.i.d.

07/2004 – Danazol use is discontinued due to liver toxicity.

07/2004 – Starts treatment with rituximab 375 mg/m2, weekly, for 4 weeks.

11/2004 – Renal biopsy

12/2004 – Starts on cyclophosphamide 100 mg/day, orally for persistent hemolysis. Hemolysis gradually resolves.

03/2006 – Cyclophosphamide is discontinued.

04/2008 – Patient remains in clinical remission.

DISCUSSION

Warm autoimmune hemolytic anemia is usually associated with extravascular hemolysis.^12–14 Hemoglobin released during extravascular hemolysis is efficiently metabolized such that free hemoglobin neither accumulates in plasma, nor appears in urine, in appreciable amounts; hemoglobinuria is thus distinctly unusual following extravascular hemolysis. This case of warm warm autoimmune hemolytic anemia is thus notable in that it provoked intravascular hemolysis which was sustained and severe, as indicated by severe anemia, significant spherocytosis, markedly reduced serum haptoglobin levels, increased LDH levels, hemoglobinuria, and hemosiderinuria.

Intravascular hemolysis releases hemoglobin into plasma where hemoglobin rapidly forms a complex with haptoglobin.³ When plasma haptoglobin is consumed because of persistent intravascular hemolysis, free hemoglobin accumulates in plasma; hemoglobin then traverses the glomerular filtration barrier and appears in the urinary space. Substantial quantities of filtered hemoglobin are endocytosed by the renal proximal tubules after which heme is freed from globin. Large amounts of free heme in cells can be cytotoxic¹⁴, and the dominant mechanism that metabolizes heme involves degradation of heme by HO;⁵ iron, released as heme is degraded by HO, is sequestered in ferritin (the major iron-storage protein) and, as occurs in certain iron overload states, in hemosiderin⁵.

To determine whether in this patient the kidney adapted to this exposure to hemoglobin, renal expression of HO-1 and ferritin was assessed. Immunoperoxidase staining demonstrated that HO-1 and ferritin were both prominently induced in the renal tubular epithelium. Induction of HO-1 and ferritin in the kidney exposed to heme proteins was first demonstrated in 1992 in a rodent model of hemolysis and rhabdomyolysis induced by the intramuscular injection of hypertonic glycerol.⁴ That such induction of HO-1 and ferritin conferred a protective response in this model was attested to by functional studies: inhibiting HO worsened AKI, whereas the prior induction of HO in this model was markedly protective against AKI and mortality.⁴ Additionally, renal ferritin content increased as HO-1 was induced in the heme protein-exposed kidney, and was dependent on the prevailing level of HO activity: renal ferritin content was reduced when HO was inhibited, and conversely, ferritin content increased when HO was induced.⁴ These findings along with studies demonstrating that increased ferritin content can bind iron and prevent iron-induced cytotoxic injury¹⁵ led to the conclusion that the coupled induction of HO-1 and ferritin provided a protective response for the following reasons: i) HO degrades heme, a potent prooxidant and potential toxicant, and ii) the increased content of ferritin binds iron released as heme is degraded, thereby avoiding iron-catalyzed oxidative stress and other types of cell injury. Support for these findings was provided by studies employing HO-1 “knockout” mice challenged with hemoglobin,¹¹ by studies which employed other inducers of HO-1,^16,17 and by studies demonstrating the protective effects of products of HO against heme protein-induced injury.¹⁸ A similar induction of HO-1 and ferritin along with increased deposition of iron has been described previously in a case of paroxysmal nocturnal hemoglobinuria,¹⁹ the latter diagnosis ruled out in our patient. Finally, the biologic importance of HO-1 is underscored by the occurrence of kidney and liver disease in conjunction with increased amounts of heme, and death in childhood, in a patient with homozygous deficiency of HO-1;^20,21 moreover, polymorphisms in the HO-1 gene associated with reduced inducibility of HO-1 may predispose to certain diseases.²²

Nephrotic range proteinuria occurred in this patient and merits comments. First, it is unlikely that proteinuria per se contributed to the induction of HO-1 since even high concentrations of albumin are incapable of inducing HO-1 in renal tubular epithelial cells.^23,24 Moreover, tubular induction of HO-1 does not occur in a model of massive, glomerular proteinuria and one which exhibits a size selective defect.²⁴ Second, such proteinuria likely reflects podocyte foot process effacement observed on renal biopsy. As this patient had longstanding hemolysis and persistent hemoglobinuria, we suggest that such trafficking of hemoglobin across the glomerular filtration barrier may injure podocyte foot processes, and thereby impair glomerular permselectivity. Indeed, in relevant disease models, the recurrent exposure of the kidney to heme proteins can induce proteinuria.²⁵ This case report thus documents that nephrotic range proteinuria can occur in chronic hemoglobinuric hemolytic anemia.

In summary, we suggest that in this patient in whom there was sustained and significant intravascular hemolysis and the attendant delivery of large amounts of hemoglobin to the kidney, the induction of HO-1 and ferritin represents a countervailing, cytoprotective renal response that mitigates the severity of AKI that may otherwise occur.