Calcineurin inhibitors (CI) are potent immunosuppressive agents used for the prevention of organ rejection and graft-versus-host disease (GVHD) after allogeneic haematopoietic stem cell transplantation (HSCT) (Shah, 1999; Trullemans et al, 2001). Both cyclosporin A (CSA) and tacrolimus (FK-506), although quite different in structure and source of extraction, belong to the class of CI and share selective antilymphocyte activity (Bartynski et al, 2001; Teive et al, 2001). CI are generally well-tolerated after transplant (Furukawa et al, 2001; Gonzalez et al, 2001; Teksam et al, 2001), but can adversely affect the skin (hypertrichosis), oral mucosa (gingival hyperplasia), cardiovascular system (hypertension) and the kidneys (elevation of creatinine) (Furlong et al, 2000; Mori et al, 2000; Shbarou et al, 2000). Other side-effects associated with CI include hyperbilirubinaemia, osteoporosis, opportunistic infections, Epstein–Barr virus-induced B-cell lymphomas and central nervous system (CNS) neurotoxicity (Zimmer et al, 1998; Coley et al, 1999; Debaere et al, 1999; Plosker & Foster, 2000). CI-induced neurotoxicity after allografting was reported in 3–20% of patients and can occur at any time after transplant. There is no clear relationship between the CI serum level and neurotoxicity (Edwards et al, 1996; Bartynski et al, 1997; Grive et al, 1997), but risk factors have been studied extensively and include hypocholestraemia, hypomagnesaemia, hypertension, acute renal failure, corticosteroid therapy, unrelated donor or human leucocyte antigen (HLA) mismatch transplants (Pettitt & Clark, 1994; Edwards et al, 1995; Pace et al, 1995), conditioning with etoposide or total-body irradiation, acute GVHD and post-transplant microangiopathic haemolytic anaemia (Labar et al, 1986; Reece et al, 1991; Fiorani et al, 1994). It is well recognized that CI-induced CNS toxicities often resolve clinically and radiologically with discontinuation of the offending drug; however, the long-term outcome of these patients is unclear. We therefore performed a retrospective analysis to determine and describe these outcomes.
Summary. Calcineurin inhibitor-induced central nervous system toxicities are uncommon and often resolve after discontinuation of the offending drug. The long-term outcome of these patients is, however, unknown. Resolution of symptoms occurred in 70% of 30 allografted recipients who developed calcineurin inhibitor-induced neurotoxicity. When patients were rechallenged with the same or a different calcineurin inhibitor, symptoms recurred in 41%, leading to permanent discontinuation of the drug. De novo or progressive acute graft-versus-host disease (GVHD) was observed in 54% of patients at a median of 7 d (range 1–70 d) after initial onset of neurotoxicity. The prognosis was grim, with 24 (80%) of these patients dying a median 33 d after the onset of neurotoxicity (range 2–594 d). GVHD and/or infection occurred in 54% and were the most common primary causes of death. We conclude that calcineurin inhibitor-induced neurotoxicity is frequently reversible but associated with a poor prognosis.
Materials and methods
A computerized search was performed for allogenic HSCT recipients who underwent magnetic resonance imaging of the brain (brain MRI) between 1996 and 2002 at Washington University. The radiological reports were browsed for the keywords ‘cyclosporine neurotoxicity’ and ‘posterior leucoencephalopathy syndrome’. Patients presenting with typical CI-induced neurotoxicity brain MRI findings, i.e. symmetric, non-enhancing, hypodense lesions on T1-weighted images with typical hyperintense T2 uptake and/or brief clinical history suggesting CI-induced neurotoxicity were identified, and medical records were reviewed. Neurotoxicity was defined as a constellation of well-characterized neurological symptoms occurring after the introduction of a CI (Fiorani et al, 1994). Patients were eligible for the present analysis if radiological and/or clinical presentation was typical of CI neurotoxicity. Epidemiological, clinical, laboratory and MRI data as well as outcomes were recorded on case report forms. Outcome data included recurrence of neurotoxicity after a switch between the two commercially available CI (CSA and FK-506), incidence and severity of acute and chronic GVHD after neurotoxicity, infectious complications, disease-free and overall survival. Results are reported as a proportion. Survival was estimated using the Kaplan–Meier product limit method.
Between January 1996 and July 2002, 542 allogenic HSCT were performed at Washington University in St Louis, and 41 patients were identified through our radiological report screening. Eleven patients were excluded because of CNS aspergillosis (n = 3), CNS leukaemia (n = 3), CNS bleed (n = 1), unexplained seizures (n = 1) or insufficient medical records (n = 3).
The characteristics of the 30 patients (5·5%) eligible for this study are summarized in Table I. Patients received allogenic bone marrow (n = 20) or peripheral blood stem cells (n = 10) from an HLA-identical sibling (n = 12), matched unrelated donor (n = 15), mismatched sibling (n = 2) or mismatched unrelated donor (n = 1) for chronic myeloid leukaemia (CML; n = 9), acute myeloid leukaemia (AML; n = 7), non-Hodgkin's lymphoma (NHL; n = 5) acute lymphocytic leukaemia (ALL; n = 2), myelodysplastic syndrome (MDS; n = 4), Hodgkin's disease (HD; n = 1), multiple myeloma (MM; n = 1) or severe aplastic anaemia (SAA; n = 1). Eighteen were male and 12 were female, median age was 46·5 years (range 18–60 years). GVHD prophylaxis consisted of CSA/methotrexate (MTX)/prednisone (n = 16), CSA/MTX (n = 3), CSA/prednisone (n = 2) or CSA alone (n = 9). At the time of neurotoxicity, 17 had prior acute GVHD, which was active but improving in 10 patients. Nine had grade I/II, five grade III and three grade IV acute GVHD. Six patients had chronic GVHD that antedated neurotoxicity.
|Patient characteristics||n (%)|
|Median age, years (range)||46·5 (18–60)|
|Source of stem cells|
|Bone marrow||20 (67)|
|Peripheral blood||10 (33)|
|Type of transplant|
|Matched sibling||12 (40)|
|Mismatched sibling||2 (7)|
|Matched unrelated donor||15 (50)|
|Mismatched unrelated donor||1 (3)|
|Post-transplant GVHD prophylaxis*|
|CsA alone||9 (30)|
|GVHD before neurotoxicity|
|Grade I/II||9 (30)|
|Grade III||5 (17)|
|Grade IV||3 (10)|
|Chronic GVHD||6 (20)|
Characteristics of neurotoxicity
Patients receiving CSA (n = 28) or FK506 (n = 2) presented a median of 43 d after transplant (range 5–717 d), with altered mental status (70%), seizures (50%), confusion (40%), visual changes (30%) including cortical blindness, ataxia (20%), severe headaches (16%), disorientation (13%), hemiparesis and/or expressive aphasia (10%) as initial manifestations of CI-induced neurotoxicity. All patients were hospitalized. Median CSA serum levels were 462 ng/ml (range 135–1116 ng/ml). Brain MRI findings were diagnostic of CI neurotoxicity in 27 patients, while three patients had non-specific MRI abnormalities despite typical clinical presentation for CI neurotoxicity.
In addition to standard supportive care, initial management of CI-induced neurotoxicity consisted of discontinuation (n = 27), dose reduction (n = 2) or continuation at the same dose (n = 1) of the CI. Neurological symptoms resolved promptly (median 3 d; range 1–11d) and completely in 21 patients (70%), including the three patients who continued the CI, and two of the three patients who presented with atypical MRI findings. Follow-up brain MRI was obtained in 11 patients, and resolution of the radiological abnormalities correlated with clinical improvement. Nine patients (30%) had irreversible neurotoxicity despite discontinuation of the CI. Two of those nine patients underwent plasma exchange without clinical or radiological improvement.
After resolution of the signs of neurotoxicity, a different (n = 10) or the same (n = 7) CI was started in 17 patients a median of 3 d (range 1–25d) after discontinuation of the initial CI. Of those, nine tolerated the CI rechallenge well, whereas seven patients (41%) had recurrence of neurotoxicity a median of 2 d (range 0–32d) from the switch from CSA to FK506 (n = 5) or the restart of the same CI at a reduced dose (n = 2). The neurological presentation was identical to the first episode of neurotoxicity in all patients, and resolved with discontinuation of the CI in only four of those eight patients.
GVHD after neurotoxicity
After discontinuation of CI, GVHD prophylaxis consisted of dose increases of prednisone (n = 18) or administration of antithymocyte globulin (n = 1), mycophenolate mofetil (n = 3) or hydroxychloroquine (n = 1). Despite these measures, 12 of the 22 patients (54%) with neurotoxicity occurring before d + 100 developed de novo (n = 4) or relapsed (n = 8) acute GVHD (two grade II, six grade III and four grade IV) a median of 7 d (range 1–70d) after initial onset of neurotoxicity. Chronic GVHD was observed in all 10 evaluable patients a median of 284 d (range 128–849d) after discontinuation of the CI.
Twenty-four patients (80%) died a median of 33 d after neurotoxicity (range 2–594 d) (Fig 1), from progressive GVHD/infection (n = 13), relapse (n = 5), multiorgan failure/thrombotic thrombocytopenic purpura (MOF/TTP; n = 5) or veno-occlusive disease (VOD; n = 1). The cause of death for the nine patients in whom neurotoxicity did not resolve was not different: GVHD/infection (n = 4), relapse (n = 2), MOF/TTP (n = 2) or VOD (n = 1). At the time of writing, six patients were still alive with a median follow-up of 6 months (range 1–41 months) and were on immunosuppression for active chronic GVHD. Table II summarizes the outcomes of patients with CI-induced neurotoxicity.
|Resolution of neurotoxicity||21 (70%)|
|Recurrence of neurotoxicity*||7 (41%)|
|Acute GVHD after neurotoxicity||12 (54%)|
|Progressive GVHD/infection||13 (54%)|
In this single-centre study, we sought to determine the long-term outcomes of allogeneic HSCT recipients who experienced neurotoxic events after administration of CI. Patients were identified through a screening of brain MRI radiological reports. This approach was considered the best available method, taking into account the specificity of typical MRI changes. Indeed, the occipital white matter is particularly susceptible to the neurotoxic effects of CI, and the detection of T2-weighted hyperintense lesions at the grey–white matter junction in the parietal or occipital lobes is typical of and diagnostic for CI-induced neurotoxicity (Shah, 1999; Bartynski et al, 2001; Trullemans et al, 2001). Patients with non-specific MRI findings who had a typical neurological syndrome were also included in this analysis, but accounted for only three cases. Although effective in identifying patients with significant neurotoxicity, our screening method was not designed to identify patients with minor neurological symptoms (tremors or headaches) that usually improve with dose reduction of CI and do not necessitate a radiological evaluation (Pettitt & Clark, 1994; Pace et al, 1995; Furukawa et al, 2001; Gonzalez et al, 2001; Teive et al, 2001). Nevertheless, the incidence of neurotoxicity observed at our centre (5·5%) and detected by radiological screening is comparable with that reported previously (Labar et al, 1986; Reece et al, 1991; Fiorani et al, 1994). Presenting symptoms in our patients were typical of CI neurotoxicity, and we found no difference in the clinical presentation between patients receiving CSA and FK506, even though the number of patients receiving FK506 was very small (n = 2).
CI-induced neurotoxicity was reversible in the majority of patients (70%) after either discontinuation or dose reduction of the CI. The reintroduction of a different or the same CI was associated with reappearance of the symptoms in approximately half the patients, regardless of the type of CI restarted or the restart doses.
In addition to the impact on the patient's quality of life and psychological status, CI-induced neurotoxicity appeared perhaps to be simply a marker to identify patients with poor long-term prognosis and at high risk of transplant-related complications. Indeed, 54% of patients who experience CI neurotoxicity before transplant d + 100 experienced de novo or progressive steroid refractory acute GVHD very shortly (7 d) after the onset of neurotoxicity. Permanent discontinuation of one of the most effective immunosuppressive drugs, i.e. CI, may also be a significant contributing factor. The survival of those patients was very poor, with 80% of patients expiring a median of 1 month after neurotoxicity.
In conclusion, this is, to our knowledge, the largest report that specifically describes the long-term outcomes of allogeneic HSCT recipients who develop CI-induced neurotoxicity. This complication is associated with high risk for morbidity and mortality, mainly from GVHD. Novel approaches for preventing GVHD after CI-induced neurotoxicity are warranted.
We thank Stephen Rodewald for the radiological computerized search, and Denize Turnbough for secretarial support.