The potential of FK506 to cause injury to blood vessels was the subject of several investigations during the developmental phases of this drug. Vascular necrosis described as vasculitis involving medium-sized arteries in the liver, pancreas, and heart was reported in FK506-treated dogs at the University of Cambridge (23
). Work performed at two other laboratories raised doubts about the significance of these findings because vasculitis was found with equal frequency in control animals (17
). Investigators in other studies were unable to reproduce vasculitis lesions in rats, baboons, or monkeys treated with FK506 (16
). Arteriolar-sized renal vessels in rats treated with FK506 can develop focal medial necrosis, accumulation of eosinophilic inclusions, and juxtaglomerular transformation, but not true arteritis (11
). These noninflammatory lesions may be the result not of a drug-induced vasculitis but of arteriolar vasospasm similar to that reported with dopaminergic and adrenergic drugs (10
). In human beings, focal necrosis of the pancreatic and peripancreatic arteries was first noted at autopsy in patients dying after organ transplantation (1
). The concomitant presence of acute pancreatitis rendered an etiologic relationship with FK506 therapy unlikely. Subsequently, the Japanese FK506 study group demonstrated arteriolar fibrinoid necrosis and glomerular thrombi in human renal allograft biopsy samples obtained from patients maintained on FK506 (7
). An improvement in serum creatinine after reduction in drug dosage was shown in only one of the reported cases. No information on plasma FK506 levels was provided.
The present study describes 10 patients with FK506-associated changes at the level of arterioles and glomerular capillaries, and correlates these changes with drug dosage, sequential plasma or whole blood FK506 levels, and serum creatinine levels. Such correlation is needed to provide firm evidence that FK506 can be toxic to the renal microvasculature. Plasma FK506 levels in cases 1–7 and 10 were indeed elevated at the time of initial kidney biopsy. Reduction of FK506 dosage led to lower drug levels and serum creatinine levels in cases 1–6 and 10, providing a clear diagnosis of FK506 toxicity. Case 7 showed no decrease in creatinine, but there was resolution of the glomerular thrombi on repeat biopsy.
Cases 8 and 9 had plasma FK506 levels within the normal range at presentation. In case 8, reduction in drug dosage was followed by a decrease in serum creatinine from 4.0 to 1.7 mg/dl, strongly suggesting drug toxicity. In case 9, serial biopsies showed the glomerular thrombi to persist for several weeks, during which time the creatinine fluctuated between 1.6 and 2.1 mg/dl. After maintaining a constant dose of FK506 for about 4 weeks, the drug intake was reduced from 10 to 8 mg/day. This was followed by resolution of the thrombi over the next 3–4 weeks. No definite improvement in the serum creatinine was observed. However, there was no clinical or histologic evidence of hypertension, acute rejection, or recurrent glomerulonephritis. The late onset of renal dysfunction (240 days after transplantation) excluded the possibility of ischemic/harvesting injury as being the cause of the glomerular capillary damage. Hence, this case also was felt to be an example of drug toxicity. The lack of elevated plasma FK506 levels is not inconsistent with this impression because correlation between drug levels and clinical events in renal transplantation is not always perfect (2
Case 6 presented as acute hemolytic uremic syndrome (HUS) after cadaveric transplantation for systemic lupus erythematosus (SLE). Several alternate explanations for the observed microvascular changes need to be considered. Spontaneously occurring HUS and thrombotic thrombocytopenic purpura (TTP)-like syndromes are described in SLE (5
). Glomerular thrombosis is reported in 50% of biopsies obtained from patients with proliferative lupus nephritis (9
). However, the patient under discussion had no clinical evidence for active SLE at the time HUS was documented. Antinuclear antibodies were undetectable, and serum complement levels were normal. A normal partial thromboplastin time was evidence against the presence of circulating lupus anticoagulants. Although we cannot definitely ascribe the clinical picture to FK506, no other satisfactory explanation for HUS was documented. Discontinuation of FK506 resulted in recovery over a period of several weeks. Another case of hemolytic uremic syndrome after FK506 therapy has been described in the literature (6
), but rigorous proof that it was precipitated by the drug was again not provided. It is notable that HUS also has been reported to occur with CS (15
), and these patients apparently can be safely switched to FK506 therapy without recurrence of HUS (13
The renal allograft biopsy samples evaluated in this study demonstrated a number of other morphologic changes previously reported in the setting of FK506 toxicity (4
). These changes provide further support for our belief that the microvascular pathology described in these patients is also drug mediated. Thus, isometric tubular cytoplasmic vacuolization was present in eight of 10 cases, and arteriolar myocyte vacuolization was present in 10 of 10 patients. Arteriolar hyalinosis was found at the time of initial biopsy in two patients (cases 3 and 4) and on follow-up biopsies in two additional patients (cases 5 and 9). Blood pressure was within the normal range in these patients, although maintenance antihypertensive therapy was needed in 2 of 4 cases. Donor disease was a potential contributing factor to the hyaline change in case 3, and diabetes was a complicating element in case 5. However, in both instances a nodular configuration to the hyaline deposits was observed similar to that previously reported in CS-associated arteriolopathy (15
). It is conceivable that the hyaline change observed in these vessels is drug induced and an aftermath of the acute microvascular injury documented earlier in the clinical course of these patients.
The spectrum of histopathological changes observed in these biopsy samples is similar to that reported with CS (15
). CS and FK506 are structurally unrelated compounds and bind to different cytosolic proteins in target cells. Nonetheless, both drugs have a closely-related mechanism of action, resulting ultimately from a block in the transcription of interleukin-2 and other cytokine genes (28
). Given these basic similarities in the immunosuppressive action of FK506 and CS, the overlap in their nephrotoxicity profile is not surprising. The actual mechanism of the microvascular injury described in this study is unknown but may be related to FK506-induced vasospasm (12
). Alternately, it may reflect direct endothelial injury and thrombosis secondary to alterations in the thromboxane A2
In summary, the data presented indicate that FK506 can cause microvascular injury in human renal allografts. The frequency of this occurrence is ~1% of renal transplant recipients if selection criteria similar to those described in this study are used. By way of comparison, fibrin thrombi are reported in 3% of diagnostic kidney biopsies obtained from CS-treated patients (15
). Glomerular capillary and arteriolar thrombi fibrin also can be seen in biopsy samples showing ischemic injury, acute rejection with intimal arteritis, sepsis with disseminated intravascular coagulation, hemolytic uremic syndrome, recurrent lupus erythematosus, and malignant hypertension. Therefore, a diagnosis of FK506-induced microvascular toxicity should be made only after careful clinicopathologic correlation and reasonable exclusion of other cases of small vessel injury.